Using the config.toml File

Admins can edit a config.toml file when starting the Driverless AI Docker image. The config.toml file includes all possible configuration options that would otherwise be specified in the nvidia-docker run command. This file is located in a folder on the container. You can make updates to environment variables directly in this file. Driverless AI will use the updated config.toml file when starting from native installs. Docker users can specify that updated config.toml file when starting Driverless AI Docker image.

Docker Image Users

  1. Copy the config.toml file from inside the Docker image to your local filesystem.
# Make a config directory
mkdir config

# Copy the config.toml file to the new config directory.
nvidia-docker run \
  --pid=host \
  --rm \
  --init \
  -u `id -u`:`id -g` \
  -v `pwd`/config:/config \
  --entrypoint bash \
  h2oai/dai-centos7-x86_64:TAG
  -c "cp /etc/dai/config.toml /config"
  1. Edit the desired variables in the config.toml file. Save your changes when you are done.
  2. Start Driverless AI with the DRIVERLESS_AI_CONFIG_FILE environment variable. Make sure this points to the location of the edited config.toml file so that the software finds the configuration file.
nvidia-docker run \
  --pid=host \
  --init \
  --rm \
  --shm-size=256m \
  -u `id -u`:`id -g` \
  -p 12345:12345 \
  -e DRIVERLESS_AI_CONFIG_FILE="/config/config.toml" \
  -v `pwd`/config:/config \
  -v `pwd`/data:/data \
  -v `pwd`/log:/log \
  -v `pwd`/license:/license \
  -v `pwd`/tmp:/tmp \
  h2oai/dai-centos7-x86_64:TAG

Native Install Users

Native installs include DEBs, RPMs, and TAR SH installs.

  1. Export the Driverless AI config.toml file or add it to ~/.bashrc. For example:
export DRIVERLESS_AI_CONFIG_FILE=“/config/config.toml”
  1. Edit the desired variables in the config.toml file. Save your changes when you are done.
  2. Start Driverless AI. Note that the command used to start Driverless AI varies depending on your install type.

For reference, below is a copy of the standard config.toml file included with this version of Driverless AI. The sections that follow describe some examples showing how to set different environment variables, data connectors, authentication, and notifications.

Sample Config.toml File

##############################################################################
##                        DRIVERLESS AI CONFIGURATION FILE
#
# Comments:
# This file is authored in TOML (see https://github.com/toml-lang/toml)
#
# Config Override Chain
# Configuration variables for Driverless AI can be provided in several ways,
# the config engine reads and overides variables in the following order
#
# 1. h2oai/config/config.toml
# [internal not visible to users]
#
# 2. config.toml
# [place file in a folder/mount file in docker container and provide path
# in "DRIVERLESS_AI_CONFIG_FILE" environment variable]
#
# 3. Environment variable
# [configuration variables can also be provided as environment variables
# they must have the prefix "DRIVERLESS_AI_" followed by
# variable name in caps e.g "authentication_method" can be provided as
# "DRIVERLESS_AI_AUTHENTICATION_METHOD"]

# Note: All floating point values < 1.0 need to start with 0.
# E.g. max_relative_cardinality = 0.95

##############################################################################
## Toml Control : Ways to control how toml parameters are set

# Whether to allow user to change non-server toml parameters per experiment in expert page
#allow_config_overrides_in_expert_page = true

# Instructions for 'Add to config.toml via toml string' in GUI expert page
# Self-referential toml parameter, for setting any other toml parameters as string of tomls separated by \n (spaces around \n are ok).
# Useful when toml parameter is not in expert mode but want per-experiment control.
# Setting this will override all other choices.
# In expert page, each time expert options saved, the new state is set without memory of any prior settings.
# The entered item is a fully compliant toml string that would be processed directly by toml.load().
# One should include 2 double quotes around the entire setting, or double quotes need to be escaped.
# One enters into the expert page text as follows:
# e.g. enable_glm=\"off\" \n enable_xgboost=\"off\" \n enable_lightgbm=\"on\"
# e.g. ""enable_glm="off" \n enable_xgboost="off" \n enable_lightgbm="off" \n enable_tensorflow="on"""
# e.g. fixed_num_individuals=4
# e.g. params_lightgbm=\"{'objective':'poisson'}\"
# e.g. ""params_lightgbm="{'objective':'poisson'}"""
# e.g. max_cores=10 \n data_precision=\"float32\" \n max_rows_feature_evolution=50000000000 \n ensemble_accuracy_switch=11 \n feature_engineering_effort=1 \n target_transformer=\"identity\" \n tournament_feature_style_accuracy_switch=5 \n params_tensorflow=\"{'layers': [100, 100, 100, 100, 100, 100]}\"
# e.g. ""max_cores=10 \n data_precision="float32" \n max_rows_feature_evolution=50000000000 \n ensemble_accuracy_switch=11 \n feature_engineering_effort=1 \n target_transformer="identity" \n tournament_feature_style_accuracy_switch=5 \n params_tensorflow="{'layers': [100, 100, 100, 100, 100, 100]}"""
# If you see: "toml.TomlDecodeError" then ensure toml is set correctly.
# When set in the expert page of an experiment, these changes only affect experiments and not the server
# Usually should keep this as empty string in this toml file.
#config_overrides = ''

##############################################################################
## Setup : Configure application server here (ip, ports, authentication, file
# types etc)

# IP address and port of autoviz process.
#vis_server_ip = "127.0.0.1"
#vis_server_port = 12346

# IP address and port of procsy process.
#procsy_ip = "127.0.0.1"
#procsy_port = 12347

# IP address and port of H2O instance.
#h2o_ip = "127.0.0.1"
#h2o_port = 54321

# IP address and port for Driverless AI HTTP server.
#ip = "127.0.0.1"
#port = 12345

# File upload limit (default 100GB)
#max_file_upload_size = 104857600000

# Verbosity of logging
# 0: quiet   (CRITICAL, ERROR, WARNING)
# 1: default (CRITICAL, ERROR, WARNING, INFO, DATA)
# 2: verbose (CRITICAL, ERROR, WARNING, INFO, DATA, DEBUG)
# Affects server and all experiments
#log_level = 1

# Whether to collect relevant server logs (h2oai_server.log, dai.log from systemctl or docker, and h2o log)
# Useful for when sending logs to H2O.ai
#collect_server_logs_in_experiment_logs = false

# Redis
#redis_ip = "127.0.0.1"
#redis_port = 6379
#master_redis_password = ""

# https settings
#
# You can make a self-signed certificate for testing with the following commands:
#
#     sudo openssl req -x509 -newkey rsa:4096 -keyout private_key.pem -out cert.pem -days 3650 -nodes -subj "/O=Driverless AI"
#     sudo chown dai:dai cert.pem private_key.pem
#     sudo chmod 600 cert.pem private_key.pem
#     sudo mv cert.pem private_key.pem /etc/dai
#
#enable_https = false
#ssl_key_file = "/etc/dai/private_key.pem"
#ssl_crt_file = "/etc/dai/cert.pem"

# SSL TLS
#ssl_no_sslv2 = true
#ssl_no_sslv3 = true
#ssl_no_tlsv1 = true
#ssl_no_tlsv1_1 = true
#ssl_no_tlsv1_2 = false
#ssl_no_tlsv1_3 = false

# Data directory. All application data and files related datasets and
# experiments are stored in this directory.

#data_directory = "./tmp"

# Whether to run quick performance benchmark at start of application
#enable_benchmark = true

# Whether to run quick startup checks at start of application
#enable_startup_checks = true

# Whether to opt in to usage statistics and bug reporting
#usage_stats_opt_in = false

# authentication_method
# unvalidated : Accepts user id and password. Does not validate password.
# none: Does not ask for user id or password. Authenticated as admin.
# openid: Users OpenID Connect provider for authentication. See additional OpenID settings below.
# pam: Accepts user id and password. Validates user with operating system.
# ldap: Accepts user id and password. Validates against an ldap server. Look
# for additional settings under LDAP settings.
# local: Accepts a user id and password. Validated against an htpasswd file provided in local_htpasswd_file.
# ibm_spectrum_conductor: Authenticate with IBM conductor auth api.
#authentication_method = "unvalidated"

# default amount of time in hours before we force user to login again (if not provided by authentication_method)
#authentication_default_timeout_hours = 72

# OpenID Connect Settings:
# base server uri to the OpenID Provider server (ex: http://localhost:7777)
#auth_openid_provider_base_uri=""
# uri to pull OpenID config data from (you can extract most of required OpenID config from this url)
# usually located at: /auth/realms/master/.well-known/openid-configuration
#auth_openid_configuration_uri=""
# uri to start authentication flow
#auth_openid_auth_uri=""
# uri to make request for token after callback from OpenID server was received
#auth_openid_token_uri=""
# uri to get user information once access_token has been acquired (ex: list of groups user belongs to will be provided here)
#auth_openid_userinfo_uri=""
# uri to logout user
#auth_openid_logout_uri=""
# callback uri that OpenID provide will use to send "authentication_code"
#auth_openid_redirect_uri=""
# OAuth2 grant type (usually acceess_token)
#auth_openid_grant_type=""
# OAuth2 response type (usually code)
#auth_openid_response_type=""
# Client ID registered with OpenID provider
#auth_openid_client_id=""
# Client secret provided by OpenID provider when registering Client ID
#auth_openid_client_secret=""
# Scope of info (usually openid)
#auth_openid_scope=""
# What key in user_info json should we check to authorize user
#auth_openid_userinfo_auth_key=""
# What value should the key have in user_info json in order to authorize user
#auth_openid_userinfo_auth_value=""
# Key that specifies username in user_info json (we will use it as username in our system)
#auth_openid_userinfo_username_key=""

# server_cookie_expiration_days
# Sets the time until expiration of secure cookie issued by server to client
# Cookie is issued upon login to Driverless AI UI and will expire 'n' days after that point
# If you wish cookies to expire in less than 1 day use decimals (1 day / 24 hours = 0.042 day/hour --> cookie expires in 1 hour)
# Expected behavior: if user is logged in, and cookie expires, the next click will redirect the user to the login page of the Driverless AI UI.
#server_cookie_expiration_days = 30

# LDAP Configuration
#ldap_server = ""    # ldap server domain or ip
#ldap_port = ""      # ldap server port
#ldap_bind_dn = ""  # Complete DN of the LDAP bind user
#ldap_bind_password = "" #Password for the LDAP bind
#ldap_tls_file = ""        # Provide Cert file location
#ldap_use_ssl = ""         # use true to use ssl or false
#ldap_search_base = ""     # the location in the DIT where the search will start
#ldap_search_filter = ""   # a string that describes what you are searching for
#ldap_search_attributes = "" # ldap attributes to return from search
#ldap_user_name_attribute ="uid" # specify key to find user name

# LDAP depricated settings
#ldap_recipe = "0"          # When using this recipe, needs to be set to "1"
#ldap_user_prefix = ""  # Deprecated do not use
#ldap_search_user_id = ""  # Depricated, Use ldap_bind_dn
#ldap_search_password = "" # Depricated, ldap_bind_password
#ldap_ou_dn = ""           # Deprecated, use ldap_search_base instead
#ldap_dc = ""               # Deprecated, use ldap_base_dn
#ldap_base_dn = ""           # Deprecated, use ldap_search_base
#ldap_base_filter = ""       # Deprecatedm use ldap_search_filter

# Local password file
# Generating a htpasswd file: see syntax below
# htpasswd -B "<location_to_place_htpasswd_file>" "<username>"
# note: -B forces use of brcypt, a secure encryption method
#local_htpasswd_file = ""

# Supported file formats (file name endings must match for files to show up in file browser)
#supported_file_types = "csv, tsv, txt, dat, tgz, gz, bz2, zip, xz, xls, xlsx, nff, jay, feather, bin, arff, parquet"

# File System Support
# upload : standard upload feature
# file : local file system/server file system
# hdfs : Hadoop file system, remember to configure the HDFS config folder path and keytab below
# dtap : Blue Data Tap file system, remember to configure the DTap section below
# s3 : Amazon S3, optionally configure secret and access key below
# gcs : Google Cloud Storage, remember to configure gcs_path_to_service_account_json below
# gbq : Google Big Query, remember to configure gcs_path_to_service_account_json below
# minio : Minio Cloud Storage, remember to configure secret and access key below
# snow : Snowflake Data Warehouse, remember to configure Snowflake credentials below (account name, username, password)
# kdb : KDB+ Time Series Database, remember to configure KDB credentials below (hostname and port, optionally: username, password, classpath, and jvm_args)
# azrbs : Azure Blob Storage, remember to configure Azure credentials below (account name, account key)
#enabled_file_systems = "upload, file, hdfs, s3"

# do_not_log_list : add configurations that you do not wish to be recorded in logs here

#do_not_log_list = "local_htpasswd_file, aws_access_key_id, aws_secret_access_key, snowflake_password, snowflake_url, snowflake_user, snowflake_account, minio_endpoint_url, minio_access_key_id, minio_secret_access_key, kdb_user, kdb_password, ldap_bind_password, gcs_path_to_service_account_json, azure_blob_account_name, azure_blob_account_key, deployment_aws_access_key_id, deployment_aws_secret_access_key, master_minio_access_key_id, master_minio_secret_access_key, master_redis_password, auth_openid_client_id, auth_openid_client_secret, auth_openid_userinfo_auth_key, auth_openid_userinfo_auth_value, auth_openid_userinfo_username_key"

# Minio is used for file distribution on multinode architecture
# These settings are used to specify the local Minio connection to master nodes
#master_minio_address = "<URL>:<PORT>"
#master_minio_access_key_id = ""
#master_minio_secret_access_key = ""

#allow_localstorage = true

##############################################################################
## Scoring Artifacts: Setup which scoring artifacts to generate by default

# Whether to create the Python scoring pipeline at the end of each experiment
#make_python_scoring_pipeline = true

# Whether to create the MOJO scoring pipeline at the end of each experiment
# Note: Not all transformers or main models are available for MOJO (e.g. no gblinear main model)
#make_mojo_scoring_pipeline = false


##############################################################################
## Hardware: Configure hardware settings here (GPUs, CPUs, Memory, etc.)

# Max number of CPU cores to use per experiment. Set to <= 0 to use all cores.
# One can also set environment variable "OMP_NUM_THREADS" to number of cores to use for OpenMP
# e.g. In bash: export OMP_NUM_THREADS=32 and export OPENBLAS_NUM_THREADS=32
#Set to -1 for all available cores.
#max_cores = -1

# Whether to set automatic number of cores by physical (true) or logical (false) count
# Using all logical cores can lead to poor performance due to cache thrashing
#max_cores_by_physical = true

# Absolute limit to core count
#max_cores_limit = 100

# Number of GPUs to use per experiment for training task.  Set to -1 for all GPUs.
# An experiment will generate many different models.
# Currently num_gpus_per_experiment!=-1 disables GPU locking, so is only recommended for
# single experiments and single users.
# Ignored if GPUs disabled or no GPUs on system.
# More info at: https://github.com/NVIDIA/nvidia-docker/wiki/nvidia-docker#gpu-isolation
#num_gpus_per_experiment = -1

# Number of GPUs to use per model training task.  Set to -1 for all GPUs.
# For example, when this is set to -1 and there are 4 GPUs available, all of them can be used for the training of a single model.
# Currently num_gpus_per_model!=1 disables GPU locking, so is only recommended for single
# experiments and single users.
# Ignored if GPUs disabled or no GPUs on system.
# More info at: https://github.com/NVIDIA/nvidia-docker/wiki/nvidia-docker#gpu-isolation
#num_gpus_per_model = 1

# Minimum number of threads for datatable (and OpenMP) during data munging
# datatable is the main data munging tool used within Driverless ai (source :
# https://github.com/h2oai/datatable)
#min_dt_threads_munging = 4

# Like min_datatable (and OpenMP)_threads_munging but for final pipeline munging
#min_dt_threads_final_munging = 4

# Which gpu_id to start with
# If using CUDA_VISIBLE_DEVICES=... to control GPUs (preferred method), gpu_id=0 is the
# first in that restricted list of devices.
# E.g. if CUDA_VISIBLE_DEVICES="4,5" then gpu_id_start=0 will refer to the
# device #4.
# E.g. from expert mode, to run 2 experiments, each on a distinct GPU out of 2 GPUs:
# Experiment#1: num_gpus_per_model=1, num_gpus_per_experiment=1, gpu_id_start=0
# Experiment#2: num_gpus_per_model=1, num_gpus_per_experiment=1, gpu_id_start=1
# E.g. from expert mode, to run 2 experiments, each on a distinct GPU out of 8 GPUs:
# Experiment#1: num_gpus_per_model=1, num_gpus_per_experiment=4, gpu_id_start=0
# Experiment#2: num_gpus_per_model=1, num_gpus_per_experiment=4, gpu_id_start=4
# E.g. Like just above, but now run on all 4 GPUs/model
# Experiment#1: num_gpus_per_model=4, num_gpus_per_experiment=4, gpu_id_start=0
# Experiment#2: num_gpus_per_model=4, num_gpus_per_experiment=4, gpu_id_start=4
# If num_gpus_per_model!=1, global GPU locking is disabled
# (because underlying algorithms don't support arbitrary gpu ids, only sequential ids),
# so must setup above correctly to avoid overlap across all experiments by all users
# More info at: https://github.com/NVIDIA/nvidia-docker/wiki/nvidia-docker#gpu-isolation
# Note that gpu selection does not wrap, so gpu_id_start + num_gpus_per_model must be less than number of visibile gpus
#gpu_id_start = 0

# Maximum number of workers for DriverlessAI server pool (only 1 needed
# currently)
#max_workers = 1

# Period (in seconds) of ping by DriverlessAI server to each experiment
# (in order to get logger info like disk space and memory usage)
# 0 means don't print anything
#ping_period = 60

# Minimum amount of disk space in GB needed to run experiments.
# Experiments will fail if this limit is crossed.
# This limit exists, because Driverless AI needs to generate data for model training
# feature engineering, documentation and other such processes.
# Disk space can also be used as an alternative to system memory (RAM).
#disk_limit_gb = 5

# Minimum amount of system memory in GB needed to start experiments
# Similarly with disk space, a certain amount of system memory is needed to run some basic
# operations.
#memory_limit_gb = 5

# Minimum number of rows needed to run experiments (values lower than 100
# might not work)
# A minimum threshold is set to ensure there is enough data to create a statistically
# reliable model and avoid other small-data related failures.
#min_num_rows = 100

# Minimum required number of rows (in the training data) for each class label for
# classification problems.
#min_rows_per_class = 5

# Minimum required number of rows for each split when generating validation samples.
#min_rows_per_split = 5

# Precision of how data is stored
# "float32" best for speed, "float64" best for accuracy or very large input values
# "float32" allows numbers up to about +-3E38 with relative error of about 1E-7
# "float64" allows numbers up to about +-1E308 with relative error of about 1E-16
# Some calculations, like the GLM standardization, can only handle up to sqrt() of these maximums for data values,
# So GLM with 32-bit precision can only handle up to about a value of 1E19 before standardization generates inf values.
# If you see "Best individual has invalid score" you may require higher precision.
#data_precision = "float32"

# Precision of most data transformers
# (Same options and notes as data_precision)
# Useful for higher precision in transformers with numerous operations that can accumulate error
# Also useful if want faster performance for transformers but otherwise want data stored in high precision
#transformer_precision = "float32"

# Whether to change ulimit soft limits up to hard limits (for DAI server app, which is not a generic user app)
# Prevents resource limit problems in some cases
# Restricted to no more than limit_nofile and limit_nproc for those resources
#ulimit_up_to_hard_limit = true

# number of file limit
# Below should be consistent with start-dai.sh
#limit_nofile=65535

# number of threads limit
# Below should be consistent with start-dai.sh
#limit_nproc=16384

##############################################################################
## Machine Learning: Configure machine learning configurations here
# (Data, Feature Engineering, Modelling etc)

# Seed for random number generator to make experiments reproducible (on same hardware), only active if 'reproducible' mode is enabled
#seed = 1234

# List of values that should be interpreted as missing values during data import. Applies both to numeric and string columns. Note that 'nan' is always interpreted as a missing value for numeric columns.
#missing_values = "['', '?', 'None', 'nan', 'NA', 'N/A', 'unknown', 'inf', '-inf', '1.7976931348623157e+308', '-1.7976931348623157e+308']"

# For tensorflow, what numerical value to give to missing values, where numeric values are standardized
# So 0 is center of distribution, and if Normal distribution then +-5 is 5 standard deviations away from the center.
# In many cases, an out of bounds value is a good way to represent missings, but in some cases the mean (0) may be better.
#tf_nan_impute_value = -5

# Internal threshold for number of rows x number of columns to trigger certain statistical
# techniques (small data recipe like including one hot encoding for all model types, and smaller learning rate)
# to increase model accuracy
#statistical_threshold_data_size_small = 100000

# Internal threshold for number of rows x number of columns to trigger certain statistical
# techniques (fewer genes created, removal of high max_depth for tree models, etc.) that can speed up modeling
# Also controls maximum rows used in training final model,
# by sampling statistical_threshold_data_size_large / columns number of rows
#statistical_threshold_data_size_large = 1000000000

# Internal threshold for number of rows x number of columns to trigger sampling for auxiliary data uses,
# like imbalanced data set detection and bootstrap scoring sample size and iterations
#aux_threshold_data_size_large = 10000000

# Internal threshold for number of rows x number of columns to trigger certain changes in performance
# (fewer threads if beyond large value) to help avoid OOM or unnecessary slowdowns
# (fewer threads if lower than small value) to avoid excess forking of tasks
#performance_threshold_data_size_small = 100000
#performance_threshold_data_size_large = 100000000

# Upper limit on the number of rows x number of columns for feature evolution (applies to both training and validation/holdout splits)
# feature evolution is the process that determines which features will be derived
# Depending on accuracy settings, a fraction of this value will be used
#feature_evolution_data_size = 100000000

# Maximum number of columns to start an experiment. This threshold exists to constraint the # complexity and the length of the Driverless AI's processes.
#max_cols = 10000

# Largest number of rows to use for column stats, otherwise sample randomly
#max_rows_col_stats = 1000000

# Maximum number of columns selected out of original set of original columns, using feature selection
# The selection is based upon how well target encoding (or frequency encoding if not available) on categoricals and numerics treated as categoricals
# This is useful to reduce the final model complexity. First the best
# [max_orig_cols_selected] are found through feature selection methods and then
# these features are used in feature evolution (to derive other features) and in modelling.
#max_orig_cols_selected = 10000

# Maximum number of numeric columns selected, above which will do feature selection
# same as above (max_orig_cols_selected) but for numeric columns.
#max_orig_numeric_cols_selected = 10000

# Maximum number of non-numeric columns selected, above which will do feature selection on all features and avoid num_as_cat
# same as above (max_orig_numeric_cols_selected) but for categorical columns.
#max_orig_nonnumeric_cols_selected = 500

# The factor times max_orig_cols_selected, by which column selection is based upon no target encoding and no num_as_cat
# in order to limit performance cost of feature engineering
#max_orig_cols_selected_simple_factor = 2

# Maximum allowed fraction of unique values for integer and categorical columns (otherwise will treat column as ID and drop)
#max_relative_cardinality = 0.95

# Maximum allowed number of unique values for integer and categorical columns (otherwise will treat column as ID and drop)
#max_absolute_cardinality = 1000000

# Whether to treat some numerical features as categorical
# For instance, sometimes an integer column may not represent a numerical feature but
# represent different numerical codes instead.
#num_as_cat = true

# Max number of unique values for integer/real columns to be treated as categoricals (test applies to first statistical_threshold_data_size_small rows only)
#max_int_as_cat_uniques = 50

# Number of folds for models used during the feature engineering process
# Increasing this will put a lower fraction of data into validation and more into training
# E.g. num_folds=3 means 67%/33% training/validation splits
# Actual value will vary for small or big data cases
#num_folds = 3

# Accuracy setting equal and above which enables full cross-validation (multiple folds) during feature evolution
# as opposed to only a single holdout split (e.g. 2/3 train and 1/3 validation holdout)
#full_cv_accuracy_switch = 8

# Accuracy setting equal and above which enables stacked ensemble as final model
# Stacking commences at the end of the feature evolution process.
# It quite often leads to better model performance, but it does increase the complexity
# and execution time of the final model.
#ensemble_accuracy_switch = 5

# Fixed ensemble_level
# -1 = auto, based upon ensemble_accuracy_switch, accuracy, size of data, etc.
# 0 = No ensemble, only final single model on validated iteration/tree count
# 1 = 1 model, multiple ensemble folds (cross-validation)
# 2 = 2 models, multiple ensemble folds (cross-validation)
# 3 = 3 models, multiple ensemble folds (cross-validation)
# 4 = 4 models, multiple ensemble folds (cross-validation)
#fixed_ensemble_level = -1

# Number of fold splits to use for ensemble_level >= 2
# The ensemble modelling may require predictions to be made on out-of-fold samples
# hence the data needs to be split on different folds to generate these predictions.
# Less folds (like 2 or 3) normally create more stable models, but may be less accurate
# More folds can get to higher accuracy at the expense of more time, but the performance
# may be less stable when the training data is not enough (i.e. higher chance of overfitting).
# Actual value will vary for small or big data cases
#num_ensemble_folds = 5

# Number of repeats for each fold for all validation
# (modified slightly for small or big data cases)
#fold_reps = 1

# For binary classification: ratio of majority to minority class equal and above which to enable undersampling
# This option helps to deal with imbalance (on the target variable)
#imbalance_ratio_undersampling_threshold = 5

# Quantile-based sampling method for imbalanced binary classification (only if class ratio is above the threshold provided above)
# Model on data is used to create deciles of predictions, and then each decile is sampled from uniformly.
#quantile_imbalanced_sampling = false

# Maximum number of classes to allow for a multi-classification problem.
# High number of classes may make certain processes of Driverless AI time-consuming.
# Memory requirements also increase with higher number of classes
#max_num_classes = 100

# Number of actuals vs. predicted data points to use in order to generate in the relevant
# plot/graph which is shown at the right part of the screen within an experiment.
#num_actuals_vs_predicted = 100

# Whether to use H2O.ai brain, the local caching and smart re-use of prior models to generate features for new models
# This variable essentially controls how much information we store about the different
# models generated and different features explored while running an experiment. It can help # with checkpointing and retrieving experiments that have been paused or interrupted.
#  Will use H2O.ai brain cache if cache file has no extra column names per column type,
#  cache exactly matches classes, class labels, and time series options,
#  interpretability of cache is equal or lower,
#  main model (booster) is allowed by new experiment
# Level of brain to use (for chosen level, where higher levels will also do all lower level operations automatically)
# -1 = Don't use any brain cache and don't write any cache
#  0 = Don't use any brain cache but still write cache
#      Use case: Want to save model for later use, but want current model to be built without any brain models
#  1 = smart checkpoint if passed in old experiment_id to pull from (via GUI, running "restart from checkpoint" or chose which experiment to resume from)
#      Use case: From GUI select prior experiment using the right-hand panel, and select "RESTART FROM LAST CHECKPOINT" to use specific experiment's model to build new models from
#  2 = smart checkpoint from H2O.ai brain cache of individual best models
#      Use case: No need to select a particular prior experiment, we scan through H2O.ai brain cache for best models to restart from
#  3 = smart checkpoint like level #1, but for entire population.  Tune only if brain population insufficient size
#      (will re-score entire population in single iteration, so appears to take longer to complete first iteration)
#  4 = smart checkpoint like level #2, but for entire population.  Tune only if brain population insufficient size
#      (will re-score entire population in single iteration, so appears to take longer to complete first iteration)
#  5 = like #4, but will scan over entire brain cache of populations to get best scored individuals, starting from resumed experiment if chosen.
#      (can be slower due to brain cache scanning if big cache)
# Other use cases:
# a) Restart on different data: Use same column names and fewer or more rows (applicable to 1 - 5)
# b) Re-fit only final pipeline: Like (a), but choose time=1 and feature_brain_level=3 - 5
# c) Restart with more columns: Add columns, so model builds upon old model built from old column names (1 - 5)
# d) Restart with focus on model tuning: Restart, then select feature_engineering_effort = 3 in expert settings
# Notes:
# 1) For Restart cases, may want to set min_dai_iterations to non-zero to force delayed early stopping, else may not be enough iterations to find better model.
# 2) A "Restart from last checkpoint" of a Re-fit will fail to find cache and re-start fresh experiment
# 3) A "New model with Same Params" of a Re-fit will fail to find cache and re-start fresh experiment
# 4) A "New model with Same Params" of a Restart will use feature_brain_level=3 for default Restart mode (revert to 2, or even 0 if want to start a fresh experiment otherwise)
#feature_brain_level = 2

# Maximum number of brain individuals pulled from H2O.ai brain cache for feature_brain_level=1, 2
#max_num_brain_indivs = 3

# Directory, relative to data_directory, to store H2O.ai brain meta model files
#brain_rel_dir = "H2O.ai_brain"

# Maximum size in bytes the brain will store
# We reserve this memory to save data in order to ensure we can retrieve an experiment if
# for any reason it gets interrupted.
# -1: unlimited
# >=0 number of GB to limit brain to
#brain_max_size_GB = 20

# Whether to enable early stopping
# Early stopping refers to stopping the feature evolution/engineering process
# when there is no performance uplift after a certain number of iterations.
# After early stopping has been triggered, Driverless AI will initiate the ensemble
# process if selected.
#early_stopping = true

# Minimum number of Driverless AI iterations to stop the feature evolution/engineering
# process even if score is not improving. Driverless AI needs to run for at least that many
# iterations before deciding to stop. It can be seen a safeguard against suboptimal (early)
# convergence.
#min_dai_iterations = 0

# Maximum features per model (and each model within the final model if ensemble) kept just after scoring them
# Keeps top varaible importance features, prunes rest away, after each scoring.
# Final ensemble will exclude any pruned-away features and only train on kept features,
#   but may contain a few new features due to fitting on different data view (e.g. new clusters)
# Final scoring pipeline will exclude any pruned-away features,
#   but may contain a few new features due to fitting on different data view (e.g. new clusters)
# -1 means no restrictions except internally-determined memory restrictions
#nfeatures_max = -1

# Recipe type
# Recipes override any GUI settings
# 'auto' : all models and features automatically determined by experiment settings, toml settings, and feature_engineering_effort
# 'compliant' : like 'auto' except:
#
# interpretability=10 (to avoid complexity, overrides GUI or python client chose for interpretability)
# enable_glm='on' (rest 'off', to avoid complexity and be compatible with algorithms supported by MLI)
# num_as_cat=false: don't convert any numerics to categoricals except via one-hot encoding (to avoid complexity)
# fixed_ensemble_level=0: Don't use any ensemble (to avoid complexity)
# feature_brain_level=0: No feature brain used (to ensure every restart is identical)
# max_feature_interaction_depth=1: interaction depth is set to 1 (no multi-feature interactions to avoid complexity)
# target_transformer='identity': for regression (to avoid complexity)
# check_distribution_shift=false: Don't use distribution shift between train, valid, and test to drop features (bit risky without fine-tuning)
#recipe = 'auto'

# How much effort to spend on feature engineering (0...10)
# Heuristic combination of various developer-level toml parameters
# 0   : keep only numeric features, only model tuning during evolution
# 1   : keep only numeric features and frequency-encoded categoricals, only model tuning during evolution
# 2   : Like #1 but instead just no Text features.  Some feature tuning before evolution.
# 3   : Like #5 but only tuning during evolution.  Mixed tuning of features and model parameters.
# 4   : Like #5, but slightly more focused on model tuning
# 5   : Default.  Balanced feature-model tuning
# 6-7 : Like #5, but slightly more focused on feature engineering
# 8   : Like #6-7, but even more focused on feature engineering with high feature generation rate, no feature dropping even if high interpretability
# 9-10: Like #8, but no model tuning during feature evolution
#feature_engineering_effort = 5

# Threshold for average string-is-text score as determined by internal heuristics
# It decides when a string column will be treated as text (for an NLP problem) or just as
# a standard categorical variable.
# Higher values will favor string columns as categoricals, lower values will favor string columns as text
#string_col_as_text_threshold = 0.3

# Mininum fraction of unique values for string columns to be considered as possible text (otherwise categorical)
#string_col_as_text_min_relative_cardinality = 0.1

# Mininum number of uniques for string columns to be considered as possible text (otherwise categorical)
#string_col_as_text_min_absolute_cardinality = 100

# Interpretability setting equal and above which will use monotonicity constraints in GBM
# You may read the following source to understand what these constraints connote and why
# they may be important, especially when the end goal is a very interpretable machine
# learning model: https://blog.datadive.net/monotonicity-constraints-in-machine-learning/
#monotonicity_constraints_interpretability_switch = 7

# Exploring feature interactions can be important in gaining better predictive performance.
# The interaction can take multiple forms (i.e. feature1 + feature2 or feature1 * feature2 + ... featureN)
# Although certain machine learning algorithms (like tree-based methods) can do well in
#capturing these interactions as part of their training process, still generating them may
# help them (or other algorithms) yield better performance.
# The depth of the interaction level (as in "u"p to"" how many features may be combined at
# once to create one single feature) can be specified to control the complexity of the
# feature engineering process. Higher values might be able to make more predictive models
# # at the expense of time.
#max_feature_interaction_depth = 8

# Accuracy setting equal and above which enables tuning of model parameters
# Only applicable if parameter_tuning_num_models=-1 (auto)
#tune_parameters_accuracy_switch = 3

# Number of models to tune during pre-evolution phase
# Can make this lower to avoid excessive tuning, or make higher to do enhanced tuning
# -1 : auto
#parameter_tuning_num_models = -1

# Accuracy setting equal and above which enables tuning of target transform for regression
# This is useful for time series when instead of predicting the actual target value, it
# might be better to predict a transformed target variable like sqrt(target) or log(target)
# as a means to control for outliers.
#tune_target_transform_accuracy_switch = 3

# Whether to automatically select target transformation for regression problems
# Can choose: 'identity' to disable any transformation, otherwise use 'auto'
#target_transformer = 'auto'

# Tournament style (method to decide which models are best at each iteration)
# "auto" : Choose based upon accuracy, etc.
# "fullstack" : Choose among optimal model and feature types
# "uniform" : all individuals in population compete to win as best
# "model" : individuals with same model type compete
# "feature" : individuals with similar feature types compete
# "model" and "feature" styles preserve at least one winner for each type (and so 2 total indivs of each type after mutation)
# For each case, a round robin approach is used to choose best scores among type of models to choose from
#tournament_style = "auto"

# Interpretability above which will use "uniform" tournament style
#tournament_uniform_style_interpretability_switch = 6

# Accuracy below which will use uniform style if tournament_style = "auto" (regardless of other accuracy tournament style switch values)
#tournament_uniform_style_accuracy_switch = 6

# Accuracy equal and above which uses model style if tournament_style = "auto"
#tournament_model_style_accuracy_switch = 6

# Accuracy equal and above which uses feature style if tournament_style = "auto"
#tournament_feature_style_accuracy_switch = 7

# Accuracy equal and above which uses fullstack style if tournament_style = "auto"
#tournament_fullstack_style_accuracy_switch = 8

# Driverless AI uses a genetic algorithm (GA) to find the best features, best models and
# best hyper parameters for these models. The GA facilitates getting good results while not
# requiring torun/try every possible model/feature/parameter. This version of GA has
# reinforcement learning elements - it uses a form of exploration-exploitation to reach
# optimum solutions. This means it will capitalise on models/features/parameters that seem # to be working well and continue to exploit them even more, while allowing some room for
# trying new (and semi-random) models/features/parameters to avoid settling on a local
# minimum.
# These models/features/parameters tried are what-we-call individuals of a population. More # individuals connote more models/features/parameters to be tried and compete to find the best # ones.
#num_individuals = 2

# set fixed number of individuals (if > 0) - useful to compare different hardware configurations
#fixed_num_individuals = 0

# set fixed number of folds (if > 0) when using cross-validation. It may be useful for
# quick runs regardless of the data size
#fixed_num_folds = 0

# set fixed number of fold reps (if > 0) - useful for quick runs regardless of data
#fixed_fold_reps = 0

# set true to force only first fold for models - useful for quick runs regardless of data
#fixed_only_first_fold_model = false

# number of unique targets or folds counts after which switch to faster/simpler non-natural sorting and print outs
#sanitize_natural_sort_limit = 1000

# Whether target encoding is generally enabled
# Target encoding refers to several different feature transformations (primarily focused on
# categorical data) that aim to represent the feature using information of the actual
# target variable. A simple example can be to use the mean of the target to replace each
# unique category of a categorical feature. This type of features can be very predictive,
# but are prone to overfitting and require more memory as they need to store mappings of
# the unique categories and the target values.
#enable_target_encoding = true

# Driverless AI categorises all data (feature engineering) transformers
# More information for these transformers can be viewed here:
# http://docs.h2o.ai/driverless-ai/latest-stable/docs/userguide/transformations.html
# This section allows excluding/blocking these transformations and may be useful when
# simpler (more interpretable) models are sought at the expense of accuracy.
# the interpretability setting)
# for multi-class: "['NumCatTETransformer', 'TextLinModelTransformer',
# 'FrequentTransformer', 'CVTargetEncodeF', 'ClusterDistTransformer',
# 'WeightOfEvidenceTransformer', 'TruncSVDNumTransformer', 'CVCatNumEncodeF',
# 'DatesTransformer', 'TextTransformer', 'FilterTransformer',
# 'NumToCatWoETransformer', 'NumToCatTETransformer', 'ClusterTETransformer',
# 'BulkInteractionsTransformer']"
#
# for regression/binary: "['TextTransformer', 'ClusterDistTransformer',
# 'FilterTransformer', 'TextLinModelTransformer', 'NumToCatTETransformer',
# 'DatesTransformer', 'WeightOfEvidenceTransformer', 'BulkInteractionsTransformer',
# 'FrequentTransformer', 'CVTargetEncodeF', 'NumCatTETransformer',
# 'NumToCatWoETransformer', 'TruncSVDNumTransformer', 'ClusterTETransformer',
# 'CVCatNumEncodeF']"
#
# This list appears in the experiment logs (search for "Transformers used")
# e.g. to disable all Target Encoding: exclude_transformers =
# "['NumCatTETransformer', 'CVTargetEncodeF', 'NumToCatTETransformer',
# 'ClusterTETransformer']"
#exclude_transformers = ""

# Exclude list of genes (i.e. genes (built on top of transformers) to not use,
# independent of the interpretability setting)
# Some transformers are used by multiple genes, so this allows different control over feature engineering
#
# for multi-class: "['BulkInteractionsGene', 'WeightOfEvidenceGene',
# 'NumToCatTargetEncodeSingleGene', 'FilterGene', 'TextGene', 'FrequentGene',
# 'NumToCatWeightOfEvidenceGene', 'NumToCatWeightOfEvidenceMonotonicGene', '
# CvTargetEncodeSingleGene', 'DateGene', 'NumToCatTargetEncodeMultiGene', '
# DateTimeGene', 'TextLinRegressorGene', 'ClusterIDTargetEncodeSingleGene',
# 'CvCatNumEncodeGene', 'TruncSvdNumGene', 'ClusterIDTargetEncodeMultiGene',
# 'NumCatTargetEncodeMultiGene', 'CvTargetEncodeMultiGene', 'TextLinClassifierGene',
# 'NumCatTargetEncodeSingleGene', 'ClusterDistGene']"
#
# for regression/binary: "['CvTargetEncodeSingleGene', 'NumToCatTargetEncodeSingleGene',
# 'CvCatNumEncodeGene', 'ClusterIDTargetEncodeSingleGene', 'TextLinRegressorGene',
# 'CvTargetEncodeMultiGene', 'ClusterDistGene', 'FilterGene', 'DateGene',
# 'ClusterIDTargetEncodeMultiGene', 'NumToCatTargetEncodeMultiGene',
# 'NumCatTargetEncodeMultiGene', 'TextLinClassifierGene', 'WeightOfEvidenceGene',
# 'FrequentGene', 'TruncSvdNumGene', 'BulkInteractionsGene', 'TextGene',
# 'DateTimeGene', 'NumToCatWeightOfEvidenceGene',
# 'NumToCatWeightOfEvidenceMonotonicGene', ''NumCatTargetEncodeSingleGene']"
#
# This list appears in the experiment logs (search for "Genes used")
# e.g. to disable bulk interaction gene, use:  exclude_genes =
#"['BulkInteractionsGene']"
#exclude_genes = ""

# Parameters for LightGBM to override DAI parameters
# parameters should be given as XGBoost equivalent unless unique LightGBM parameter
# e.g. 'eval_metric' instead of 'metric' should be used
# e.g. params_lightgbm = "{'objective': 'binary:logistic', 'n_estimators': 100, 'max_leaves': 64, 'random_state': 1234}"
# e.g. params_lightgbm = {'n_estimators': 600, 'learning_rate': 0.1, 'reg_alpha': 0.0, 'reg_lambda': 0.5, 'gamma': 0, 'max_depth': 0, 'max_bin': 128, 'max_leaves': 256, 'scale_pos_weight': 1.0, 'max_delta_step': 3.469919910597877, 'min_child_weight': 1, 'subsample': 0.9, 'colsample_bytree': 0.3, 'tree_method': 'gpu_hist', 'grow_policy': 'lossguide', 'min_data_in_bin': 3, 'min_child_samples': 5, 'early_stopping_rounds': 20, 'num_classes': 2, 'objective': 'binary:logistic', 'eval_metric': 'logloss', 'random_state': 987654, 'early_stopping_threshold': 0.01, 'monotonicity_constraints': False, 'silent': True, 'debug_verbose': 0, 'subsample_freq': 1}"
# avoid including "system"-level parameters like 'n_gpus': 1, 'gpu_id': 0, , 'n_jobs': 1, 'booster': 'lightgbm'
# also likely should avoid parameters like: 'objective': 'binary:logistic', unless one really knows what one is doing (e.g. alternative objectives)
# See: https://xgboost.readthedocs.io/en/latest/parameter.html
# And see: https://github.com/Microsoft/LightGBM/blob/master/docs/Parameters.rst
# Can also pass objective parameters if choose (or in case automatically chosen) certain objectives
# https://lightgbm.readthedocs.io/en/latest/Parameters.html#metric-parameters
#params_lightgbm = "{}"

# Parameters for XGBoost to override DAI parameters
# similar parameters as lightgbm since lightgbm parameters are transcribed from xgboost equivalent versions
# e.g. params_xgboost = "{'n_estimators': 100, 'max_leaves': 64, 'max_depth': 0, 'random_state': 1234}"
# See: https://xgboost.readthedocs.io/en/latest/parameter.html
#params_xgboost = "{}"

# Like params_xgboost but for XGBoost's dart method
#params_dart = "{}"

# Parameters for Tensorflow to override DAI parameters
# e.g. params_tensorflow = "{'lr': 0.01, 'add_wide': False, 'add_attention': True, 'epochs': 30, 'layers': [100, 100], 'activation': 'selu', 'batch_size': 64, 'chunk_size': 1000, 'dropout': 0.3, 'strategy': 'one_shot', 'l1': 0.0, 'l2': 0.0, 'ort_loss': 0.5, 'ort_loss_tau': 0.01, 'normalize_type': 'streaming'}"
# See: https://keras.io/ , e.g. for activations: https://keras.io/activations/
# Example layers: [500, 500, 500], [100, 100, 100], [100, 100], [50, 50]
# Strategies: '1cycle' or 'one_shot', See: https://github.com/fastai/fastai
# normalize_type: 'streaming' or 'global' (using sklearn StandardScaler)
#params_tensorflow = "{}"

# Parameters for XGBoost's gblinear to override DAI parameters
# e.g. params_gblinear = "{'n_estimators': 100}"
# See: https://xgboost.readthedocs.io/en/latest/parameter.html
#params_gblinear = "{}"

# Parameters for Rulefit to override DAI parameters
# e.g. params_rulefit = "{'max_leaves': 64}"
# See: https://xgboost.readthedocs.io/en/latest/parameter.html
#params_rulefit = "{}"

# Parameters for FTRL to override DAI parameters
#params_ftrl = "{}"

# Dictionary of key:lists of values to use for LightGBM tuning, overrides DAI's choice per key
# e.g. params_tune_lightgbm = "{'min_child_samples': [1,2,5,100,1000], 'min_data_in_bin': [1,2,3,10,100,1000]}"
#params_tune_lightgbm = "{}"

# Like params_tune_lightgbm but for XGBoost
# e.g. params_tune_xgboost = "{'max_leaves': [8, 16, 32, 64]}"
#params_tune_xgboost = "{}"

# Like params_tune_lightgbm but for XGBoost's Dart
# e.g. params_tune_dart = "{'max_leaves': [8, 16, 32, 64]}"
#params_tune_dart = "{}"

# Like params_tune_lightgbm but for TensorFlow
# e.g. params_tune_tensorflow = "{'layers': [[10,10,10], [10, 10, 10, 10]]}"
#params_tune_tensorflow = "{}"

# Like params_tune_lightgbm but for gblinear
# e.g. params_tune_gblinear = "{'reg_lambda': [.01, .001, .0001, .0002]}"
#params_tune_gblinear = "{}"

# Like params_tune_lightgbm but for rulefit
# e.g. params_tune_rulefit = "{'max_depth': [4, 5, 6]}"
#params_tune_rulefit = "{}"

# Like params_tune_lightgbm but for ftrl
#params_tune_ftrl = "{}"

# Whether to force max_leaves and max_depth to be 0 if grow_policy is depthwise and lossguide, respectively.
#params_tune_grow_policy_simple_trees = true

# Whether to enable XGBoost models (auto/on/off)
#enable_xgboost = "auto"

# Internal threshold for number of rows x number of columns to trigger no xgboost models due to high memory use
# Overridden if enable_xgboost = "on", in which case always allow xgboost to be used
#xgboost_threshold_data_size_large = 100000000

# Internal threshold for number of rows x number of columns to trigger no xgboost models due to limits on GPU memory capability
# Overridden if enable_xgboost = "on", in which case always allow xgboost to be used
#xgboost_gpu_threshold_data_size_large = 30000000

# Whether to enable GLM models (auto/on/off)
#enable_glm = "auto"

# Whether to enable LightGBM models (auto/on/off)
#enable_lightgbm = "auto"

# Whether to enable Random Forest (in LightGBM package) models (auto/on/off/only)
#enable_rf = "auto"

# Whether to enable TensorFlow models (beta version, no mojo) (auto/on/off)
#enable_tensorflow = "off"

# Whether to enable RuleFit support (beta version, no mojo) (auto/on/off)
#enable_rulefit = "off"

# Whether to enable FTRL support (beta version, no mojo) (follow the regularized leader) model (auto/on/off)
#enable_ftrl = "off"

# Maximum number of GBM trees or GLM iterations
# Early-stopping usually chooses less
#max_nestimators = 3000

# factor by which max_nestimators is reduced for tuning and feature evolution
#max_nestimators_feature_evolution_factor = 0.2

# Maximum tree depth (and corresponding max max_leaves as 2**max_max_depth)
#max_max_depth = 12

# Default max_bin for tree methods
#default_max_bin = 256

# Default max_bin for lightgbm (recommended for GPU lightgbm)
#default_lightgbm_max_bin = 64

# Maximum max_bin for any tree
#max_max_bin = 256

# Minimum max_bin for any tree
#min_max_bin = 32

# Amount of memory which can handle max_bin = 256 can handle 125 columns and max_bin = 32 for 1000 columns
# As available memory on system goes higher than this scale, can handle proportionally more columns at higher max_bin
# Currently set to 10GB
#scale_mem_for_max_bin = 10737418240

# Factor by which rf gets more depth than gbdt
#factor_rf = 1.5

# Upper limit on learning rate for GBM models
# If want to override min_learning_rate and min_learning_rate_final, set this to smaller value
#max_learning_rate = 0.5

# Lower limit on learning rate for feature engineering GBM models
#min_learning_rate = 0.05

# Lower limit on learning rate for final ensemble GBM models
#min_learning_rate_final = 0.01

# Max. number of epochs for TensorFlow models
#tensorflow_max_epochs = 10

# Whether tensorflow will use all CPU cores, or if it will split among all transformers
#tensorflow_use_all_cores = true

# Whether tensorflow will use all CPU cores if reproducible is set, or if it will split among all transformers
#tensorflow_use_all_cores_even_if_reproducible_true = false

# Max. number of epochs for TensorFlow models for making NLP features
#tensorflow_max_epochs_nlp = 2

# Accuracy setting equal and above which will add all enabled TensorFlow NLP models below at start of experiment for text dominated problems
#enable_tensorflow_nlp_accuracy_switch = 5

# Whether to use Word-based CNN TensorFlow models for NLP if tensorflow enabled
#enable_tensorflow_textcnn = false

# Whether to use Word-based Bi-GRU TensorFlow models for NLP if tensorflow enabled
#enable_tensorflow_textbigru = false

# Whether to use Character-level CNN TensorFlow models for NLP if tensorflow enabled
#enable_tensorflow_charcnn = false

# Max number of rules to be used for RuleFit models (-1 for all)
#rulefit_max_num_rules = -1

# Max tree depth for RuleFit models
#rulefit_max_tree_depth = 6

# Max number of trees for RuleFit models
#rulefit_max_num_trees = 100

# Internal threshold for number of rows x number of columns to trigger no rulefit models due to being too slow currently
#rulefit_threshold_data_size_large = 100000000

# Enable One-Hot-Encoding (which does binning to limit to number of bins to no more than 100 anyway) for categorical columns with fewer than this many unique values
# Set to 0 to disable
#one_hot_encoding_cardinality_threshold = 50

# list of possible bins for target encoding (first is default value)
#te_bin_list = [25, 10, 100, 250]

# list of possible bins for weight of evidence encoding (first is default value)
# If only want one value: woe_bin_list = [2]
#woe_bin_list = [25, 10, 100, 250]

# list of possible bins for ohe hot encoding (first is default value)
#ohe_bin_list = [10, 25, 50, 75, 100]

# Enable time series recipe
#time_series_recipe = true

# earliest datetime for automatic conversion of integers in %Y%m%d format to a time column during parsing
#min_ymd_timestamp = 19700101

# lastet datetime for automatic conversion of integers in %Y%m%d format to a time column during parsing
#max_ymd_timestamp = 20300101

# maximum number of data samples (randomly selected rows) for date/datetime format detection
#max_rows_datetime_format_detection = 100000

# Whether to enable train/valid and train/test distribution shift detection
#check_distribution_shift = true

# Normalized training variable importance above which to check the feature for shift
# Useful to avoid checking likely unimportant features
#shift_key_features_varimp = 0.01

# Whether to only check certain features based upon the value of shift_key_features_varimp
#check_reduced_features = true

# Number of trees to use to train model to check shift in distribution
# No larger than max_nestimators
#shift_trees = 100

# The value of max_bin to use for trees to use to train model to check shift in distribution
#shift_max_bin = 256

# The value of max_depth to use for trees to use to train model to check shift in distribution
#shift_max_depth = 4

# If distribution shift detection is enabled, show features for which shift AUC is above this value
# (AUC of a binary classifier that predicts whether given feature value belongs to train or test data)
#detect_features_distribution_shift_threshold_auc = 0.55

# If distribution shift detection is enabled, drop features for which shift AUC is above this value
# (AUC of a binary classifier that predicts whether given feature value belongs to train or test data)
#drop_features_distribution_shift_threshold_auc = 0.6

# Minimum number of features to keep, keeping least shifted feature at least if 1
#drop_features_distribution_shift_min_features = 1

# Whether to enable detailed traces (in GUI Trace)
#detailed_traces = false

# How close to the optimal value (usually 1 or 0) does the validation score need to be to be considered perfect (to stop the experiment)?
#abs_tol_for_perfect_score = 1e-4

#############################################################################
##Time Series settings

# Normalized probability of choosing to lag non-targets relative to targets
#prob_lag_non_targets = 0.1

# Unnormalized probability of choosing other lag time-series transformers based on interactions
#prob_lagsinteraction = 0.1

# Unnormalized probability of choosing other lag time-series transformers based on aggregations
#prob_lagsaggregates = 0.1

# Automatically generate is-holiday features from date columns
#holiday_features = true

# County code to use to look up holiday calendar (Python package 'holiday')
#holiday_country = "US"

# Max. sample size for automatic determination of time series train/valid split properties, only if time column is selected
#max_time_series_properties_sample_size = 1000000

# Maximum number of lag sizes, which are sampled from if sample_lag_sizes==true, else all are taken (-1 == automatic)
#max_lag_sizes = -1

# Minimum required autocorrelation threshold for a lag to be considered for feature engineering
#min_lag_autocorrelation = 0.1

# How many samples of lag sizes to use for a single time group (single time series signal)
#max_signal_lag_sizes = 100

# Whether to sample lag sizes
#sample_lag_sizes = false

# Probability for new Lags/EWMA gene to use default lags (determined by frequency/gap/horizon, independent on data)
#prob_default_lags = 0.2

# How many samples of lag sizes to use, chosen randomly out of original set of lag sizes
#max_sampled_lag_sizes = 10

# Override lags to be used
# e.g. [7, 14, 21] # this exact list
# e.g. 21 # produce from 1 to 21
# e.g. 21:3 produce from 1 to 21 in step of 3
# e.g. 5-21 produce from 5 to 21
# e.g. 5-21:3 produce from 5 to 21 in step of 3
#override_lag_sizes = []

# Whether to consider time groups columns as potential features
#allow_tgc_memorization = false

# Maximum time t spent to generate training holdout predictions
# t < 0: up to 1440 minutes (24h) spent for generating training holdout predictions
# t = 0: no training holdout predictions
# t > 0: up to t minutes spent for generating training holdout predictions
#time_series_holdout_predictions_timebank = 0

##################################################################################
##MLI ( Machine Learning Intepretability) settings
# When number of rows are above this limit sample for MLI for scoring UI data
#mli_sample_above_for_scoring = 1000000

# When number of rows are above this limit sample for MLI for training surrogate models
#mli_sample_above_for_training = 100000

# When sample for MLI how many rows to sample
#mli_sample_size = 100000

# how many bins to do quantile binning
#mli_num_quantiles = 10

# mli random forest number of trees
#mli_drf_num_trees = 100

# Whether to speed up predictions used inside MLI with a fast approximation
#mli_fast_approx = true

# mli number of trees for fast_approx during predict for Shapley
#fast_approx_num_trees = 50

# whether to do only 1 fold and 1 model of all folds and models if ensemble
#fast_approx_do_one_fold_one_model = true

# mli random forest max depth
#mli_drf_max_depth = 20

# not only sample training, but also sample scoring
#mli_sample_training = true

# regularization strength for k-LIME GLM's
#klime_lambda = [1e-6, 1e-8]
#klime_alpha = 0.0

# mli converts numeric columns to enum when cardinality is <= this value
#mli_max_numeric_enum_cardinality = 25

# Maximum number of features allowed for k-LIME k-means clustering
#mli_max_number_cluster_vars = 6

#Use all columns for k-LIME k-means clustering (this will override `mli_max_number_cluster_vars` if set to `true`
#use_all_columns_klime_kmeans = false

#Strict version check for MLI
#mli_strict_version_check = true

#MLI cloud name
#mli_cloud_name = ""

##############################################################################
## Machine Learning Output : What kinds of files are written related to the machine learning process

# Whether to dump every scored individual's variable importance (both derived and original) to csv/tabulated/json file
# produces files like: individual_scored_id%d.iter%d*features*
#dump_varimp_every_scored_indiv = false

# Whether to dump every scored individual's model parameters to csv/tabulated file
# produces files like: individual_scored_id%d.iter%d*params*
#dump_modelparams_every_scored_indiv = false

# Whether to append (false) or have separate files (true) for modelparams every scored indiv
#dump_modelparams_separate_files = false

# Location of the AutoDoc template
#autodoc_template = "report_template.docx"

# Whether to compute training, validation, and test correlation matrix (table and heatmap pdf) and save to disk
# alpha: currently single threaded and slow for many columns
#compute_correlation = false

# Whether to dump to disk a correlation heatmap
#produce_correlation_heatmap = false

# Value to report high correlation between original features
#high_correlation_value_to_report = 0.95

# Whether to dump to *timings.txt files timing for each transformer
#write_trans_timings = true

# whether to delete preview timings if wrote transformer timings
#delete_preview_trans_timings = true

# Whether to delete preview cache on server exit
#preview_cache_upon_server_exit = true


##############################################################################
## Connectors : Configure connector specifications here
#
# Instance Local file system
# The option disable access to DAI data_directory from file browser
#file_hide_data_directory = true
# The option specify include only list of absolute path prefixes
# which will be only accessible in file browser.
# For example:
# file_path_filter_include = "['/data','/home/michal/']"
#file_path_filter_include = "[]"

# Enable usage of path filters
#file_path_filtering_enabled = false

## HDFS
## Note that if using Kerberos, be sure that the DAI time
## is synched with the Kerberos server.

# Configurations for a HDFS data source
# Path of hdfs coresite.xml
# core_site_xml_path is deprecated, please use hdfs_config_path
#core_site_xml_path = ""

# HDFS config folder path , can contain multiple config files
#hdfs_config_path = ""

# Path of the principal key tab file
#key_tab_path = ""

# HDFS connector
# Auth type can be Principal/keytab/keytabPrincipal
# Specify HDFS Auth Type, allowed options are:
#   noauth : No authentication needed
#   principal : Authenticate with HDFS with a principal user
#   keytab : Authenticate with a Key tab (recommended). If running
#             DAI as a service, then the Kerberos keytab needs to
#             be owned by the DAI user.
#   keytabimpersonation : Login with impersonation using a keytab
#hdfs_auth_type = "noauth"

# Kerberos app principal user (recommended)
#hdfs_app_principal_user = ""

# Deprecated: Do not use hdfs_app_login_user, user name is taken from user login
#hdfs_app_login_user = ""

# JVM args for HDFS distributions, provide args seperate by space
# -Djava.security.krb5.conf=<path>/krb5.conf
# -Dsun.security.krb5.debug=true
# -Dlog4j.configuration=file:///<path>log4j.properties
#hdfs_app_jvm_args = ""
# hdfs class path
#hdfs_app_classpath = ""
# Limit files returned from HDFS
#hdfs_max_files_listed = 100

# Blue Data DTap connector settings are similar to HDFS connector settings.
#
# Specify DTap Auth Type, allowed options are:
#   noauth : No authentication needed
#   principal : Authenticate with DTab with a principal user
#   keytab : Authenticate with a Key tab (recommended). If running
#             DAI as a service, then the Kerberos keytab needs to
#             be owned by the DAI user.
#   keytabimpersonation : Login with impersonation using a keytab
#
# NOTE: "hdfs_app_classpath" and "core_site_xml_path" are both required to be set for DTap connector
#dtap_auth_type = "noauth"
# Dtap (HDFS) config folder path , can contain multiple config files
#dtap_config_path = ""
# Path of the principal key tab file
#dtap_key_tab_path = ""
# Kerberos app principal user (recommended)
#dtap_app_principal_user = ""
# Specify the user id of the current user here as user@realm
#dtap_app_login_user = ""
# JVM args for DTap distributions, provide args seperate by space
#dtap_app_jvm_args = ""
# DTap (HDFS) class path. NOTE: set "hdfs_app_classpath" also
#dtap_app_classpath = ""

# S3 Connector credentials
#aws_access_key_id = ""
#aws_secret_access_key = ""
#aws_role_arn = ""

# What region to use when none is specified in the s3 url.
# Ignored when aws_s3_endpoint_url is set.
#aws_default_region = ""

# Sets enpoint URL that will be used to access S3.
#aws_s3_endpoint_url = ""

# If set to true S3 Connector will try to to obtain credentials assiciated with
# the role attached to the EC2 instance.
#aws_use_ec2_role_credentials = false

# Starting S3 path displayed in UI S3 browser
#s3_init_path = "s3://h2o-public-test-data/smalldata/"

# GCS Connector credentials
# example (suggested) -- "/licenses/my_service_account_json.json"
#gcs_path_to_service_account_json = ""

# Minio Connector credentials
#minio_endpoint_url = ""
#minio_access_key_id = ""
#minio_secret_access_key = ""

# Snowflake Connector credentials
# Recommended Provide: url, user, password
# Optionally Provide: account, user, password
# Example URL: https://<snowflake_account>.<region>.snowflakecomputing.com
#snowflake_url = ""
#snowflake_user = ""
#snowflake_password = ""
#snowflake_account = ""

# KDB Connector credentials
#kdb_user = ""
#kdb_password = ""
#kdb_hostname = ""
#kdb_port = ""
#kdb_app_classpath = ""
#kdb_app_jvm_args = ""

# Azure Blob Store Connector credentials
#azure_blob_account_name = ""
#azure_blob_account_key = ""
#azure_connection_string = ""

# Notification scripts
# - the variable points to a location of script which is executed at given event in experiment lifecycle
# - the script should have executable flag enabled
# - use of absolute path is suggested
# The on experiment start notification script location
#listeners_experiment_start = ""
# The on experiment finished notification script location
#listeners_experiment_done = ""

# Default AWS credentials to be used for scorer deployments.
#deployment_aws_access_key_id = ""
#deployment_aws_secret_access_key = ""
#deployment_aws_bucket_name = ""

# Allow the browser to store e.g. login credentials in login form (set to false for higher security)
#allow_form_autocomplete = true

# Enable Projects workspace (alpha version, for evalulation)
#enable_projects = false

##############################################################################
## END