/Projects_Portfolio

(Old) Here are my projects related to Data Analysis and Machine Learning.

Primary LanguageJupyter Notebook

Projects_Portfolio

Here are my projects related to Data Analysis and Machine Learning.

Machine Learning Competitions (Kaggle / Zindi)

  • 1- Financial Inclusion in Africa by IndabaX Sudan 2021:

In this binary classification task, One should implement a model to predict whether a person has a bank account or not given some information about the person.

Platform: Zindi

Type of Data: Tabular Data

Position: 1 out of 38 contestants

MAE: 0.109555087371421

Contest URL: https://zindi.africa/hackathons/indabax-sudan-2021

Tools: Jupyter Notebooks - Python 3 - XGBoost - Pandas

  • 2- Titanic Survival Prediction Challenge:

In this binary classification task, One should implement a model to predict whether a person has survived from the sinking of the Titanic or not given some information about the person.

Platform: Kaggle

Type of Data: Tabular Data

Accuracy Score: 78.4%

Contest URL: https://www.kaggle.com/c/titanic/overview

Tools: Jupyter Notebooks - Python 3 - Logistic Regression - Pandas


  • 3- Economic Well-Being Prediction Challenge:

In this regression task, One should implement a model to predict a measure of wealth for different locations across Africa.

Platform: Zindi

Type of Data: Tabular Data

Position: 2 out of 107 contestants

RMSE: 0.1039661172059555

Contest URL: https://zindi.africa/competitions/umojahack-2022-practice-challenge

Tools: Jupyter Notebooks - Python 3 - LGBM - Bagging Regressor - Pandas


  • 4- Faulty Air Quality Sensor Prediction:

In this binary classification task, One should implement a model to predict whether a device is faulty based on its output data.

Platform: Zindi

Type of Data: Time Series Data

[Original Solution] Accuracy Score: 83.4%
[Updated Solution] Accuracy Score: 94.7%

[Original Solution] Position: 46 out of 719 contestants
[Updated Solution] Position: 1 out of 719 contestants

Contest URL: https://zindi.africa/competitions/umojahack-africa-2022-beginner-challenge

Tools: Jupyter Notebooks - Python 3 - LGBM - SMOTE - Pandas


  • 5- Tanzania Tourism Prediction Challenge:

In this regression task, One should implement a model to predict what a tourist will spend when visiting Tanzania.

Platform: Zindi

Type of Data: Tabular Data

Position: 13 out of 308 contestants

MAE: 4971847.829238593

Contest URL: https://zindi.africa/competitions/tanzania-tourism-prediction

Tools: Jupyter Notebooks - Python 3 - LGBM - Bagging Regressor - Pandas


  • 6- Rossmann Stores Sales Prediction:

In this regression task, One should implement a model to forecast the store sales for the next six weeks of the data.

Platform: Kaggle

Type of Data: Time Series Data

RMSPE Score: 0.11323

Contest URL: https://www.kaggle.com/competitions/rossmann-store-sales/

Tools: Jupyter Notebooks - EDA - LGBM - Stacking - Pandas


  • 7- Laduma Analytics Football League Winners Prediction:

In this multiclass classification task, One should implement a model to predict the outcome of a football match, based on historical match and player data.

Platform: Zindi

Type of Data: Time Series Data

Accuracy Score: 47.8%

Position: 4 out of 251 contestants

Contest URL: https://zindi.africa/competitions/zindi-weekendz-laduma-analytics-football-league-winners-prediction-challenge

Tools: Jupyter Notebooks - Python 3 - LGBM - Pandas


  • 8- Layer.ai Air Quality Prediction:

In this regression task, One should implement a model to predict air quality in regions in Kampala, based on satellite radar data from Sentinel 5P.

Platform: Zindi

Type of Data: Time Series Data

MAE Score: 13.33173593948973

Position: 3 out of 153 contestants

Contest URL: https://zindi.africa/competitions/zindi-weekendz-layerai-air-quality-prediction-challenge/

Tools: Jupyter Notebooks - Python 3 - CatBoost - Pandas


  • 9- AI4D Africa’s Anglophone Research Lab Tanzania Tourism Prediction:

In this multiclass classification task, One should implement a model to classify the range of expenditures a tourist spends in Tanzania.

Platform: Zindi

Type of Data: Tabular Data

Position: 32 out of 461 contestants

Log-Loss: 1.040574812022115

Contest URL: https://zindi.africa/competitions/ai4d-lab-tanzania-tourism-classification-challenge/

Tools: Jupyter Notebooks - Python 3 - Ensembling - Voting Classifier - Pandas


  • 10- Alvin Smart Money Management Classification Challenge:

In this multiclass classification task, One should implement a model to classify purchases into 13 different categories, based on transactions data from Alvin.

Platform: Zindi

Type of Data: Tabular Data

Logloss Score: 1.372343425651699

Position: 8 out of 448 contestants

Contest URL: https://zindi.africa/competitions/alvin-smart-money-management-classification-challenge/

Tools: Jupyter Notebooks - Python 3 - CatBoost - Random Forest - KNN - Pandas


  • 11- American Express Default Prediction Challenge:

In this binary classification task, One should implement a model to predict the probability that a customer does not pay back their credit card balance amount in the future based on their monthly customer profile.

Platform: Kaggle

Type of Data: Tabular Data

AMEX Score: 0.80712

Position: 450 out of 4875 contestants (A Bronze Medal)

Contest URL: https://www.kaggle.com/competitions/amex-default-prediction/

Tools: Jupyter Notebooks - Python 3 - CatBoost - CUDF - CUPY


  • 12- Fossil Demand Forecasting Challenge:

In this regression task, One should implement a model to determine as accurately as possible the demand for each individual product, four months into the future.

Platform: Zindi

Type of Data: Time Series Data

MAE Score: 165985.6352266829

Position: 24 out of 827 contestants

Contest URL: https://zindi.africa/competitions/fossil-stock-forecasting-challenge/

Tools: Jupyter Notebooks - Python 3 - CatBoost - Pandas


  • 13- Electricity and Gas Consumption Challenge:

In this classification task, One should implement a model to detect and recognize clients involved in fraudulent activities.

Platform: Zindi

Type of Data: Tabular Data

AUC Score: 0.8966153

Position: 1 out of 51 contestants

Contest URL: https://zindi.africa/competitions/indabax-sudan-classification2022/

Tools: Jupyter Notebooks - Python 3 - CatBoost - Pandas