Add a new tabular dataset for clustering problems inside tab_automl/datasets and also add the dataset class inside tab_automl/automl/datasets.py.

Add proper comments and quality in code.

Follow contributing guidelines on README.md

The PR fix for issues #7

Added KNN classifier to models.py the advantage of this is it takes almost zero time to train because it only stores the data of the training part. and faster than all the models mentioned in the models.py file , it is also a non parametric models with only parameter that needs to be mentioned is the number of neighbours, adding to this as KNN doesnt undergo training we can add new data to it which doesn't affect the accuracy of the model. It is also very easy to implement and interpret as there is only one hyperparameter which is the number of neighbours .

PR tagged with #32

I have added 5 clustering models in automl -> models.py . Name of the clustering models are as follows: AffinityPropagation AgglomerativeClustering Birch DBSCAN KMeans . DESCRIPTION AffinityPropagation - It involves finding a set of exemplars that best summarize the data. It takes as input measures of similarity between pairs of data points. Real-valued messages are exchanged between data points until a high-quality set of exemplars and corresponding clusters gradually emerges . AgglomerativeClustering- It involves merging examples until the desired number of clusters is achieved.It is implemented via the class AgglomerativeClustering and the main configuration to tune is the “n_clusters” set, an estimate of the number of clusters in the data, e.g. 2. . BIRCH -BIRCH Clustering involves constructing a tree structure from which cluster centroids are extracted. main configuration to tune is the “threshold” and “n_clusters” hyperparameters, the latter of which provides an estimate of the number of clusters. . DBSCAN Clustering involves finding high-density areas in the domain and expanding those areas of the feature space around them as clusters. the main configuration to tune is the “eps” and “min_samples” hyperparameters. . KMEANS -the main configuration to tune is the “n_clusters” hyperparameter set to the estimated number of clusters in the data.

PR fix for issue #6

Added KNN regressor to models.py the advantage of this is it takes almost zero time to train because it only stores the data of the training part. and faster than all the models mentioned in the models.py file in training time as its zero in KNN regressor , it is also a non parametric models with only parameter that needs to be mentioned is the number of neighbours, adding to this as KNN doesn't undergo training we can add new data to it which doesn't affect the accuracy of the model. It is also very easy to implement and interpret as there is only one hyperparameter which is the number of neighbours, apart from this it's versatile and can be used as a regressor as well as classifier

PR Fix for issue #3

I have updated all the Stings in print statements to f-string. I have changed tab_automl/automl/datasets.py, tab_automl/automl/fet_engineering.py, tab_automl/automl/processing.py, tab_automl/automl/training.py, tab_automl/main.py, test.py files and updated all the print statements in these files.

new file formats which are added are: .txt .json .xlsx .sqlite

changes are made on ClassificationDataset in datasets.py

One input is added to get table name from the sqlite in the .sqlite file format

The datasets are getting loaded on .csv format only in the codebase, example : see here.

Add different data loading techniques for other formats like .txt , .sqlite, etc. Add required comments in the code.

Follow contributing guidelines on README.md

This PR fixes for issue #32

Changes made

Added 5 new clustering models in models.py

Reason

Mini-Batch K-Means Mini-Batch K-Means is a modified version of k-means that makes updates to the cluster centroids using mini-batches of samples rather than the entire dataset, which can make it faster for large datasets, and perhaps more robust to statistical noise. Mean Shift Mean shift clustering involves finding and adapting centroids based on the density of examples in the feature space. OPTICS OPTICS clustering (where OPTICS is short for Ordering Points To Identify the Clustering Structure) is a modified version of DBSCAN. Spectral Clustering Spectral Clustering is a general class of clustering methods, drawn from linear algebra. to tune is the n_clusters hyperparameter used to specify the estimated number of clusters in the data. Gaussian Mixture Model A Gaussian mixture model summarizes a multivariate probability density function with a mixture of Gaussian probability distributions.

This PR fixes for issue #25

Changes made:

• Added new clustering dataset Credit Card Customer data to datasets
• Updated datasets,py with clustering class clustering() and clustering dataset class called Credit_Card_Customer_Data().

The PR fix for issues #7

Added KNN classifier to models.py the advantage of this is it takes almost zero time to train because it only stores the data of the training part. and faster than all the models mentioned in the models.py file , it is also a non parametric models with only parameter that needs to be mentioned is the number of neighbours, adding to this as KNN doesnt undergo training we can add new data to it which doesn't affect the accuracy of the model. It is also very easy to implement and interpret as there is only one hyperparameter which is the number of neighbours .

This PR fixes for issue #7

Changes made

• Added new classification model XGBoost Classifier in models.py
• Added new requirements for the model in requirements.txt

Reason

XGBoost Classifier model handles the missing data efficiently and it also has built in cross validation capability. It is also regularized, so the models don't overfit. To add, it also uses gradient descent algorithm to minimize loss. So this model can give good accuracy.

What you have to do- 1. Add 5 new clustering models inside tab_automl.automl.models file's single_model_dict object. 2. Follow same code representations.

Follow contributing guidelines on README.md

What you have to do- 1. Add a new function inside the tab_automl.automl.training.Trainer class for training clustering problems. 2. Inside the main function of tab_automl.main some implementation will be needed so that the clustering problem type fits.

Follow contributing guidelines on README.md

What you have to do - 1. Update the parser's problem type definitions. 2. Update the tab_automl.utils.misc.validate_parse_variable as it was prepared to check only the problem types of classification and regression. 3. The target variable parser should have a default value None as the clustering problem won't allow any target variable, but keep in mind if the problem type is some supervised technique, then the target_feature should be checked inside .tab_automl.utils.misc.validate_parse_variable function. 4. Also update the README.md where it specifies the problem types.

Follow contributing guidelines on README.md

1. Add k-fold validation for chosen datasets.
2. Add appropriate print statements and comments inside the code.
3. Add all utilities on tab_automl.utils.training
4. If possible update the parser too with a variable named -kf --k-fold which takes the number of folds. (Optional)

Follow contributing guidelines on README.md

1. Prepare a new class under the processing module.
2. Prepare the functions with a proper idea and also add appropriate comments.
3. Add a function "run" inside the "OutlierProcessing" which will go through every feature, e.g. link.
4. Add the function under the class Preprocessing.

Follow contributing guidelines on README.md

1. Add 3 loss functions for both regression and classification problem types.
2. Add them similarly to how the model scores are stored. See here
3. Add proper comments.
4. If new functions are needed for the loss functions, store them on tab_automl.utils.training .
5. Update the requirements if new libraries are being used.

Follow contributing guidelines on README.md