Skip to content

Latest commit

 

History

History
68 lines (49 loc) · 5.31 KB

README.md

File metadata and controls

68 lines (49 loc) · 5.31 KB
Raw

Kaggle_projects

This repo will be about implementation of supervised learning essentials on Kaggle datasets from simple models to ensemble models alongwith feature engineering and model selection. For this I will be using cleaned data acquired from Kaggle while walking you through my analysis in individual .ipynb files. Details about each dataset is provided in the notebook and below I will mention the machine learning techniques I will be progressing through. As I progress in the project number, I will include the techniques from the earlier projects, and therefore, the last project would be the most complex project for this repo. I will not demonstrate any deep learning techniques in this repo.

Also included is the link to two other Kaggle projects, one for a Regression problem and the other for a Classification problem. The objection for these two projects was to complete them within 8 hours.

Spotify Song Attributes dataset

Golden Rule, Decision Tree, data splitting, model building, hyperparameters, Dummy classification, cross validation, matplotlib

The Adult Census dataset

K Nearest Neighbors (KNN), Support Vector Machine (RBF SVM), Exploratory Data Analysis (EDA), column transformation, preprocessing, imputation, OneHotEncoder, OrdinalEncoder, StandardScaler, altair

SMS Spam Detection dataset

text classification, CountVectorizer, model selection, hyperparameter optimization , GridSearchCV, RandomizedSearchCV, pipeline, ColumnTransformer

Happy DB dataset

Naive Bayes, BernoulliNB, MultinomialNB, Logistic Regression

Churn in Telecom dataset

Random Forest,Ridge, confusion matrix, classification report, F1 score, Recall, Precision, Area Under Curve (AUC), PR curve, ROC curve, ROC AUC score, RMSE, MAPE score, make_scorer

Disaster Tweets dataset

Polynomial features, RidgeCV, RFECV, Sequential Feature Selector, Forward Selection, Backward Selection, Recursive Selection

Spotify Tracks DB dataset

SHAP, feature_importances_, Ensembles, Voting model, Stacking model, Random Forest, Gradient Boosted Trees, XGBoost, LightGBM, CatBoost,

Credit Card Default - Classification Project

ScikitLearn, pandas, Altair, SHAP, eli5, VotingClassifier, StackingClassifier, CatBoostClassifier, LGBMClassifier, Random Forest, Lasso, Support Vector Machine (SVC), CountVectorizer, RandomizedSearchCV, RFECV, cross_validate, pipeline, column_transformer, SelectFromModel, OneHotEncoder, StandardScaler, classification_report, recall_score, f1_score, precision_score, accuracy_score

Airbnb NYC - Regression Project

ScikitLearn, pandas, Altair, SHAP, eli5, CatBoostRegressor, LGBMRegressor, Random Forest, Ridge, Lasso, Suppor Vector Machine (SVR), CountVectorizer, RandomizedSearchCV, RFECV, cross_validate, pipeline, column_transformer, SelectFromModel, OneHotEncoder, StandardScaler


To use this repo

Clone this Github repository, install the dependencies, and run the following commands at the command line/terminal from the root directory of the project:

conda env create --file env.yaml
conda activate Kaggle_projects

Setting up Kaggle API

To use the Kaggle API, sign up for a Kaggle account at https://www.kaggle.com. Then go to the 'Account' tab of your user profile (https://www.kaggle.com//account) and select 'Create API Token'. This will trigger the download of kaggle.json, a file containing your API credentials. Place this file in the location ~/.kaggle/kaggle.json. I have already included the kaggle package in the repo environment, and running the below script should download the required files.

To download the data files

Run the following commands at the command line/terminal from the root directory of the project to download the data files in a /downloads folder:

make all

To reset the repo to the original state, and delete all downloaded files from Kaggle, run the following commands at the command line/terminal from the root directory of the project:

make clean

You can now run each individual notebook file.

To contribute to the repository:

  1. Fork the repository.
  2. Add the implementation of the algorithm with a clearly defined filename for the script or the notebook.
  3. Test the implementation thoroughly and make sure that it works with some dataset.
  4. Add a link with a short description about the file in the README.md.
  5. Create a pull request for review with a short description of your changes.
  6. Do not forget to add attribution for references and sources used in the implementation.