A delightful machine learning tool that allows you to train/fit, test and use models without writing code

• Free software: MIT license
• Documentation: https://igel.readthedocs.io.

The goal of the project is to provide machine learning for everyone, both technical and non-technical users.

I needed a tool sometimes, which I can use to fast create a machine learning prototype. Whether to build some proof of concept, create a fast draft model to prove a point or use auto ML. I find myself often stuck at writing boilerplate code and thinking too much where to start. Therefore, I decided to create this tool.

igel is built on top of other ML frameworks. It provides a simple way to use machine learning without writing a single line of code. Igel is highly customizable, but only if you want to. Igel does not force you to customize anything. Besides default values, igel can use auto-ml features to figure out a model that can work great with your data.

All you need is a yaml (or json) file, where you need to describe what you are trying to do. That's it!

Igel supports regression, classification and clustering. Igel's supports auto-ml features like ImageClassification and TextClassification

Igel supports most used dataset types in the data science field. For instance, your input dataset can be a csv, txt, excel sheet, json or even html file that you want to fetch. If you are using auto-ml features, then you can even feed raw data to igel and it will figure out how to deal with it. More on this later in the examples.

• Supports most dataset types (csv, txt, excel, json, html) even just raw data stored in folders
• Supports all state of the art machine learning models (even preview models)
• Supports different data preprocessing methods
• Provides flexibility and data control while writing configurations
• Supports cross validation
• Supports both hyperparameter search (version >= 0.2.8)
• Supports yaml and json format
• Usage from GUI
• Supports different sklearn metrics for regression, classification and clustering
• Supports multi-output/multi-target regression and classification
• Supports multi-processing for parallel model construction
• Support for auto machine learning

• The easiest way is to install igel using pip

$ pip install -U igel

Igel's supported models:

+--------------------+----------------------------+-------------------------+
|      regression    |        classification      |        clustering       |
+--------------------+----------------------------+-------------------------+
|   LinearRegression |         LogisticRegression |                  KMeans |
|              Lasso |                      Ridge |     AffinityPropagation |
|          LassoLars |               DecisionTree |                   Birch |
| BayesianRegression |                  ExtraTree | AgglomerativeClustering |
|    HuberRegression |               RandomForest |    FeatureAgglomeration |
|              Ridge |                 ExtraTrees |                  DBSCAN |
|  PoissonRegression |                        SVM |         MiniBatchKMeans |
|      ARDRegression |                  LinearSVM |    SpectralBiclustering |
|  TweedieRegression |                      NuSVM |    SpectralCoclustering |
| TheilSenRegression |            NearestNeighbor |      SpectralClustering |
|    GammaRegression |              NeuralNetwork |               MeanShift |
|   RANSACRegression | PassiveAgressiveClassifier |                  OPTICS |
|       DecisionTree |                 Perceptron |                KMedoids |
|          ExtraTree |               BernoulliRBM |                    ---- |
|       RandomForest |           BoltzmannMachine |                    ---- |
|         ExtraTrees |       CalibratedClassifier |                    ---- |
|                SVM |                   Adaboost |                    ---- |
|          LinearSVM |                    Bagging |                    ---- |
|              NuSVM |           GradientBoosting |                    ---- |
|    NearestNeighbor |        BernoulliNaiveBayes |                    ---- |
|      NeuralNetwork |      CategoricalNaiveBayes |                    ---- |
|         ElasticNet |       ComplementNaiveBayes |                    ---- |
|       BernoulliRBM |         GaussianNaiveBayes |                    ---- |
|   BoltzmannMachine |      MultinomialNaiveBayes |                    ---- |
|           Adaboost |                       ---- |                    ---- |
|            Bagging |                       ---- |                    ---- |
|   GradientBoosting |                       ---- |                    ---- |
+--------------------+----------------------------+-------------------------+

For auto ML:

• ImageClassifier
• TextClassifier
• ImageRegressor
• TextRegressor
• StructeredDataClassifier
• StructeredDataRegressor
• AutoModel

The help command is very useful to check supported commands and corresponding args/options

$ igel --help

You can also run help on sub-commands, for example:

$ igel fit --help

Igel is highly customizable. If you know what you want and want to configure your model manually, then check the next sections, which will guide you on how to write a yaml or a json config file. After that, you just have to tell igel, what to do and where to find your data and config file. Here is an example:

$ igel fit --data_path 'path_to_your_csv_dataset.csv' --yaml_path 'path_to_your_yaml_file.yaml'

However, you can also use the auto-ml features and let igel do everything for you. A great example for this would be image classification. Let's imagine you already have a dataset of raw images stored in a folder called images

All you have to do is run:

$ igel auto-train --data_path 'path_to_your_images_folder' --task ImageClassification

That's it! Igel will read the images from the directory, process the dataset (converting to matrices, rescale, split, etc...) and start training/optimizing a model that works good on your data. As you can see it's pretty easy, you just have to provide the path to your data and the task you want to perform.

You can run the help command to get instructions. You can also run help on sub-commands!

$ igel --help

First step is to provide a yaml file (you can also use json if you want)

You can do this manually by creating a .yaml file (called igel.yaml by convention but you can name if whatever you want) and editing it yourself. However, if you are lazy (and you probably are, like me :D), you can use the igel init command to get started fast, which will create a basic config file for you on the fly.

"""
igel init --help


Example:
If I want to use neural networks to classify whether someone is sick or not using the indian-diabetes dataset,
then I would use this command to initialize a yaml file n.b. you may need to rename outcome column in .csv to sick:

$ igel init -type "classification" -model "NeuralNetwork" -target "sick"
"""
$ igel init

After running the command, an igel.yaml file will be created for you in the current working directory. You can check it out and modify it if you want to, otherwise you can also create everything from scratch.

• Demo:

# model definition
model:
    # in the type field, you can write the type of problem you want to solve. Whether regression, classification or clustering
    # Then, provide the algorithm you want to use on the data. Here I'm using the random forest algorithm
    type: classification
    algorithm: RandomForest     # make sure you write the name of the algorithm in pascal case
    arguments:
        n_estimators: 100   # here, I set the number of estimators (or trees) to 100
        max_depth: 30       # set the max_depth of the tree

# target you want to predict
# Here, as an example, I'm using the famous indians-diabetes dataset, where I want to predict whether someone have diabetes or not.
# Depending on your data, you need to provide the target(s) you want to predict here
target:
    - sick

In the example above, I'm using random forest to classify whether someone have diabetes or not depending on some features in the dataset I used the famous indian diabetes in this example indian-diabetes dataset)

Notice that I passed n_estimators and max_depth as additional arguments to the model. If you don't provide arguments then the default will be used. You don't have to memorize the arguments for each model. You can always run igel models in your terminal, which will get you to interactive mode, where you will be prompted to enter the model you want to use and type of the problem you want to solve. Igel will then show you information about the model and a link that you can follow to see a list of available arguments and how to use these.

• The expected way to use igel is from terminal (igel CLI):

Run this command in terminal to fit/train a model, where you provide the path to your dataset and the path to the yaml file

$ igel fit --data_path 'path_to_your_csv_dataset.csv' --yaml_path 'path_to_your_yaml_file.yaml'

# or shorter

$ igel fit -dp 'path_to_your_csv_dataset.csv' -yml 'path_to_your_yaml_file.yaml'

"""
That's it. Your "trained" model can be now found in the model_results folder
(automatically created for you in your current working directory).
Furthermore, a description can be found in the description.json file inside the model_results folder.
"""

• Demo:

You can then evaluate the trained/pre-fitted model:

$ igel evaluate -dp 'path_to_your_evaluation_dataset.csv'
"""
This will automatically generate an evaluation.json file in the current directory, where all evaluation results are stored
"""

• Demo:

Finally, you can use the trained/pre-fitted model to make predictions if you are happy with the evaluation results:

$ igel predict -dp 'path_to_your_test_dataset.csv'
"""
This will generate a predictions.csv file in your current directory, where all predictions are stored in a csv file
"""

• Demo:

You can combine the train, evaluate and predict phases using one single command called experiment:

$ igel experiment -DP "path_to_train_data path_to_eval_data path_to_test_data" -yml "path_to_yaml_file"

"""
This will run fit using train_data, evaluate using eval_data and further generate predictions using the test_data
"""

• Demo:

You can export the trained/pre-fitted sklearn model into ONNX:

$ igel export -dp "path_to_pre-fitted_sklearn_model"

"""
This will convert the sklearn model into ONNX
"""

• Alternatively, you can also write code if you want to:

from igel import Igel

Igel(cmd="fit", data_path="path_to_your_dataset", yaml_path="path_to_your_yaml_file")
"""
check the examples folder for more
"""

The next step is to use your model in production. Igel helps you with this task too by providing the serve command. Running the serve command will tell igel to serve your model. Precisely, igel will automatically build a REST server and serve your model on a specific host and port, which you can configure by passing these as cli options.

The easiest way is to run:

$ igel serve --model_results_dir "path_to_model_results_directory"

Notice that igel needs the --model_results_dir or shortly -res_dir cli option in order to load the model and start the server. By default, igel will serve your model on localhost:8000, however, you can easily override this by providing a host and a port cli options.

$ igel serve --model_results_dir "path_to_model_results_directory" --host "127.0.0.1" --port 8000

Igel uses FastAPI for creating the REST server, which is a modern high performance framework and uvicorn to run it under the hood.

This example was done using a pre-trained model (created by running igel init --target sick -type classification) and the Indian Diabetes dataset under examples/data. The headers of the columns in the original CSV are ‘preg’, ‘plas’, ‘pres’, ‘skin’, ‘test’, ‘mass’, ‘pedi’ and ‘age’.

CURL:

• Post with single entry for each predictor

$ curl -X POST localhost:8080/predict --header "Content-Type:application/json" -d '{"preg": 1, "plas": 180, "pres": 50, "skin": 12, "test": 1, "mass": 456, "pedi": 0.442, "age": 50}'

Outputs: {"prediction":[[0.0]]}

• Post with multiple options for each predictor

$ curl -X POST localhost:8080/predict --header "Content-Type:application/json" -d '{"preg": [1, 6, 10], "plas":[192, 52, 180], "pres": [40, 30, 50], "skin": [25, 35, 12], "test": [0, 1, 1], "mass": [456, 123, 155], "pedi": [0.442, 0.22, 0.19], "age": [50, 40, 29]}'

Outputs: {"prediction":[[1.0],[0.0],[0.0]]}

Caveats/Limitations:

• each predictor used to train the model must make an appearance in your data (i.e. don’t leave any columns out)
• each list must have the same number of elements or you’ll get an Internal Server Error
• as an extension of this, you cannot mix single elements and lists (i.e. {“plas”: 0, “pres”: [1, 2]} isn't allowed)
• the predict function takes a data path arg and reads in the data for you but with serving and calling your served model, you’ll have to parse the data into JSON yourself however, the python client provided in examples/python_client.py will do that for you

Example usage of the Python Client:

from python_client import IgelClient

# the client allows additional args with defaults:
# scheme="http", endpoint="predict", missing_values="mean"
client = IgelClient(host='localhost', port=8080)

# you can post other types of files compatible with what Igel data reading allows
client.post("my_batch_file_for_predicting.csv")

Outputs: <Response 200>: {"prediction":[[1.0],[0.0],[0.0]]}

The main goal of igel is to provide you with a way to train/fit, evaluate and use models without writing code. Instead, all you need is to provide/describe what you want to do in a simple yaml file.

Basically, you provide description or rather configurations in the yaml file as key value pairs. Here is an overview of all supported configurations (for now):

# dataset operations
dataset:
    type: csv  # [str] -> type of your dataset
    read_data_options: # options you want to supply for reading your data (See the detailed overview about this in the next section)
        sep:  # [str] -> Delimiter to use.
        delimiter:  # [str] -> Alias for sep.
        header:     # [int, list of int] -> Row number(s) to use as the column names, and the start of the data.
        names:  # [list] -> List of column names to use
        index_col: # [int, str, list of int, list of str, False] -> Column(s) to use as the row labels of the DataFrame,
        usecols:    # [list, callable] -> Return a subset of the columns
        squeeze:    # [bool] -> If the parsed data only contains one column then return a Series.
        prefix:     # [str] -> Prefix to add to column numbers when no header, e.g. ‘X’ for X0, X1, …
        mangle_dupe_cols:   # [bool] -> Duplicate columns will be specified as ‘X’, ‘X.1’, …’X.N’, rather than ‘X’…’X’. Passing in False will cause data to be overwritten if there are duplicate names in the columns.
        dtype:  # [Type name, dict maping column name to type] -> Data type for data or columns
        engine:     # [str] -> Parser engine to use. The C engine is faster while the python engine is currently more feature-complete.
        converters: # [dict] -> Dict of functions for converting values in certain columns. Keys can either be integers or column labels.
        true_values: # [list] -> Values to consider as True.
        false_values: # [list] -> Values to consider as False.
        skipinitialspace: # [bool] -> Skip spaces after delimiter.
        skiprows: # [list-like] -> Line numbers to skip (0-indexed) or number of lines to skip (int) at the start of the file.
        skipfooter: # [int] -> Number of lines at bottom of file to skip
        nrows: # [int] -> Number of rows of file to read. Useful for reading pieces of large files.
        na_values: # [scalar, str, list, dict] ->  Additional strings to recognize as NA/NaN.
        keep_default_na: # [bool] ->  Whether or not to include the default NaN values when parsing the data.
        na_filter: # [bool] -> Detect missing value markers (empty strings and the value of na_values). In data without any NAs, passing na_filter=False can improve the performance of reading a large file.
        verbose: # [bool] -> Indicate number of NA values placed in non-numeric columns.
        skip_blank_lines: # [bool] -> If True, skip over blank lines rather than interpreting as NaN values.
        parse_dates: # [bool, list of int, list of str, list of lists, dict] ->  try parsing the dates
        infer_datetime_format: # [bool] -> If True and parse_dates is enabled, pandas will attempt to infer the format of the datetime strings in the columns, and if it can be inferred, switch to a faster method of parsing them.
        keep_date_col: # [bool] -> If True and parse_dates specifies combining multiple columns then keep the original columns.
        dayfirst: # [bool] -> DD/MM format dates, international and European format.
        cache_dates: # [bool] -> If True, use a cache of unique, converted dates to apply the datetime conversion.
        thousands: # [str] -> the thousands operator
        decimal: # [str] -> Character to recognize as decimal point (e.g. use ‘,’ for European data).
        lineterminator: # [str] -> Character to break file into lines.
        escapechar: # [str] ->  One-character string used to escape other characters.
        comment: # [str] -> Indicates remainder of line should not be parsed. If found at the beginning of a line, the line will be ignored altogether. This parameter must be a single character.
        encoding: # [str] -> Encoding to use for UTF when reading/writing (ex. ‘utf-8’).
        dialect: # [str, csv.Dialect] -> If provided, this parameter will override values (default or not) for the following parameters: delimiter, doublequote, escapechar, skipinitialspace, quotechar, and quoting
        delim_whitespace: # [bool] -> Specifies whether or not whitespace (e.g. ' ' or '    ') will be used as the sep
        low_memory: # [bool] -> Internally process the file in chunks, resulting in lower memory use while parsing, but possibly mixed type inference.
        memory_map: # [bool] -> If a filepath is provided for filepath_or_buffer, map the file object directly onto memory and access the data directly from there. Using this option can improve performance because there is no longer any I/O overhead.

    random_numbers: # random numbers options in case you wanted to generate the same random numbers on each run
        generate_reproducible:  # [bool] -> set this to true to generate reproducible results
        seed:   # [int] -> the seed number is optional. A seed will be set up for you if you didn't provide any

    split:  # split options
        test_size: 0.2  #[float] -> 0.2 means 20% for the test data, so 80% are automatically for training
        shuffle: true   # [bool] -> whether to shuffle the data before/while splitting
        stratify: None  # [list, None] -> If not None, data is split in a stratified fashion, using this as the class labels.

    preprocess: # preprocessing options
        missing_values: mean    # [str] -> other possible values: [drop, median, most_frequent, constant] check the docs for more
        encoding:
            type: oneHotEncoding  # [str] -> other possible values: [labelEncoding]
        scale:  # scaling options
            method: standard    # [str] -> standardization will scale values to have a 0 mean and 1 standard deviation  | you can also try minmax
            target: inputs  # [str] -> scale inputs. | other possible values: [outputs, all] # if you choose all then all values in the dataset will be scaled


# model definition
model:
    type: classification    # [str] -> type of the problem you want to solve. | possible values: [regression, classification, clustering]
    algorithm: NeuralNetwork    # [str (notice the pascal case)] -> which algorithm you want to use. | type igel algorithms in the Terminal to know more
    arguments:          # model arguments: you can check the available arguments for each model by running igel help in your terminal
    use_cv_estimator: false     # [bool] -> if this is true, the CV class of the specific model will be used if it is supported
    cross_validate:
        cv: # [int] -> number of kfold (default 5)
        n_jobs:   # [signed int] -> The number of CPUs to use to do the computation (default None)
        verbose: # [int] -> The verbosity level. (default 0)
    hyperparameter_search:
        method: grid_search   # method you want to use: grid_search and random_search are supported
        parameter_grid:     # put your parameters grid here that you want to use, an example is provided below
            param1: [val1, val2]
            param2: [val1, val2]
        arguments:  # additional arguments you want to provide for the hyperparameter search
            cv: 5   # number of folds
            refit: true   # whether to refit the model after the search
            return_train_score: false   # whether to return the train score
            verbose: 0      # verbosity level

# target you want to predict
target:  # list of strings: basically put here the column(s), you want to predict that exist in your csv dataset
    - put the target you want to predict here
    - you can assign many target if you are making a multioutput prediction

A detailed overview of the configurations you can provide in the yaml (or json) file is given below. Notice that you will certainly not need all the configuration values for the dataset. They are optional. Generally, igel will figure out how to read your dataset.

However, you can help it by providing extra fields using this read_data_options section. For example, one of the helpful values in my opinion is the "sep", which defines how your columns in the csv dataset are separated. Generally, csv datasets are separated by commas, which is also the default value here. However, it may be separated by a semicolon in your case.

Hence, you can provide this in the read_data_options. Just add the sep: ";" under read_data_options.

A complete end to end solution is provided in this section to prove the capabilities of igel. As explained previously, you need to create a yaml configuration file. Here is an end to end example for predicting whether someone have diabetes or not using the decision tree algorithm. The dataset can be found in the examples folder.

• Fit/Train a model:

model:
    type: classification
    algorithm: DecisionTree

target:
    - sick

$ igel fit -dp path_to_the_dataset -yml path_to_the_yaml_file

That's it, igel will now fit the model for you and save it in a model_results folder in your current directory.

• Evaluate the model:

Evaluate the pre-fitted model. Igel will load the pre-fitted model from the model_results directory and evaluate it for you. You just need to run the evaluate command and provide the path to your evaluation data.

$ igel evaluate -dp path_to_the_evaluation_dataset

That's it! Igel will evaluate the model and store statistics/results in an evaluation.json file inside the model_results folder

• Predict:

Use the pre-fitted model to predict on new data. This is done automatically by igel, you just need to provide the path to your data that you want to use prediction on.

$ igel predict -dp path_to_the_new_dataset

That's it! Igel will use the pre-fitted model to make predictions and save it in a predictions.csv file inside the model_results folder

You can also carry out some preprocessing methods or other operations by providing them in the yaml file. Here is an example, where the data is split to 80% for training and 20% for validation/testing. Also, the data are shuffled while splitting.

Furthermore, the data are preprocessed by replacing missing values with the mean ( you can also use median, mode etc..). check this link for more information

# dataset operations
dataset:
    split:
        test_size: 0.2
        shuffle: True
        stratify: default

    preprocess: # preprocessing options
        missing_values: mean    # other possible values: [drop, median, most_frequent, constant] check the docs for more
        encoding:
            type: oneHotEncoding  # other possible values: [labelEncoding]
        scale:  # scaling options
            method: standard    # standardization will scale values to have a 0 mean and 1 standard deviation  | you can also try minmax
            target: inputs  # scale inputs. | other possible values: [outputs, all] # if you choose all then all values in the dataset will be scaled

# model definition
model:
    type: classification
    algorithm: RandomForest
    arguments:
        # notice that this is the available args for the random forest model. check different available args for all supported models by running igel help
        n_estimators: 100
        max_depth: 20

# target you want to predict
target:
    - sick

Then, you can fit the model by running the igel command as shown in the other examples

$ igel fit -dp path_to_the_dataset -yml path_to_the_yaml_file

For evaluation

$ igel evaluate -dp path_to_the_evaluation_dataset

For production

$ igel predict -dp path_to_the_new_dataset

In the examples folder in the repository, you will find a data folder,where the famous indian-diabetes, iris dataset and the linnerud (from sklearn) datasets are stored. Furthermore, there are end to end examples inside each folder, where there are scripts and yaml files that will help you get started.

The indian-diabetes-example folder contains two examples to help you get started:

• The first example is using a neural network, where the configurations are stored in the neural-network.yaml file
• The second example is using a random forest, where the configurations are stored in the random-forest.yaml file

The iris-example folder contains a logistic regression example, where some preprocessing (one hot encoding) is conducted on the target column to show you more the capabilities of igel.

Furthermore, the multioutput-example contains a multioutput regression example. Finally, the cv-example contains an example using the Ridge classifier using cross validation.

You can also find a cross validation and a hyperparameter search examples in the folder.

I suggest you play around with the examples and igel cli. However, you can also directly execute the fit.py, evaluate.py and predict.py if you want to.

First, create or modify a dataset of images that are categorized into sub-folders based on the image label/class For example, if you are have dogs and cats images, then you will need 2 sub-folders:

• folder 0, which contains cats images (here the label 0 indicates a cat)
• folder 1, which contains dogs images (here the label 1 indicates a dog)

Assuming these two sub-folder are contained in one parent folder called images, just feed data to igel:

$ igel auto-train -dp ./images --task ImageClassification

Igel will handle everything from pre-processing the data to optimizing hyperparameters. At the end, the best model will be stored in the current working dir.

First, create or modify a text dataset that are categorized into sub-folders based on the text label/class For example, if you are have a text dataset of positive and negative feedbacks, then you will need 2 sub-folders:

• folder 0, which contains negative feedbacks (here the label 0 indicates a negative one)
• folder 1, which contains positive feedbacks (here the label 1 indicates a positive one)

Assuming these two sub-folder are contained in one parent folder called texts, just feed data to igel:

$ igel auto-train -dp ./texts --task TextClassification

Igel will handle everything from pre-processing the data to optimizing hyperparameters. At the end, the best model will be stored in the current working dir.

You can also run the igel UI if you are not familiar with the terminal. Just install igel on your machine as mentioned above. Then run this single command in your terminal

$ igel gui

This will open up the gui, which is very simple to use. Check examples of how the gui looks like and how to use it here: https://github.com/nidhaloff/igel-ui

• Use the official image (recommended):

You can pull the image first from docker hub

$ docker pull nidhaloff/igel

Then use it:

$ docker run -it --rm -v $(pwd):/data nidhaloff/igel fit -yml 'your_file.yaml' -dp 'your_dataset.csv'

• Alternatively, you can create your own image locally if you want:

You can run igel inside of docker by first building the image:

$ docker build -t igel .

And then running it and attaching your current directory (does not need to be the igel directory) as /data (the workdir) inside of the container:

$ docker run -it --rm -v $(pwd):/data igel fit -yml 'your_file.yaml' -dp 'your_dataset.csv'

• Article: https://medium.com/@nidhalbacc/machine-learning-without-writing-code-984b238dd890

If you are facing any problems, please feel free to open an issue. Additionally, you can make contact with the author for further information/questions.

Do you like igel? You can always help the development of this project by:

• Following on github and/or twitter
• Star the github repo
• Watch the github repo for new releases
• Tweet about the package
• Help others with issues on github
• Create issues and pull requests
• Sponsor the project

You think this project is useful and you want to bring new ideas, new features, bug fixes, extend the docs?

Contributions are always welcome. Make sure you read the guidelines first

MIT license

Copyright (c) 2020-present, Nidhal Baccouri