abdelhakiem/weatherprediction

WeatherPrediction

This project is about creating a machine learning classification model to predict weather states. The project involved several steps such as data exploration, cleaning, visualization, feature engineering, and model deployement. The following are the main stages of the project:

Data Exploration

In the first step, the weather dataset was explored. This included checking for missing data, data types, and duplicates. Any errors were fixed and the data was cleaned for the next step.

Data Visualization

After cleaning, necessary data visualizations were created. This step allowed us to better understand the relationships between different features and the target variable. We also checked for any correlation between different features and target variables.

Feature Engineering

Next, we performed feature engineering to select the necessary features. This step involved selecting the most relevant features that could be used for model building.

Model Building

In this step, we built five classification models: Naive Bayes, Logistic Regression, SVM, KNN, and XGBoost. We evaluated the performance of each model and selected the model with the highest accuracy, which was Naive Bayes accuracy = 84.7%.

Web Application

We built a data-driven web application using Streamlit, a Python web framework for building interactive web applications. The web application allows users to input the necessary weather features and get the predicted weather state.

Deployment

Finally, we deployed the model on the Streamlit-sharing website. The deployed model can be accessed through this LINK . The user interface is simple and intuitive, allowing users to easily input the necessary data and get the predicted weather state.

bandicam.2023-05-15.12-39-45-938.mp4

Conclusion

In conclusion, this project involved creating a machine learning classification model to predict weather states. We explored the data, performed data cleaning and visualization, feature engineering, and model building. We also built a data-driven web application using Streamlit and deployed the model on Streamlit-sharing website. The final model achieved an accuracy of 84.7% and can be easily accessed through the web application.