deepthisudharsan / rainfall-pattern-prediction-using-ml Goto Github PK

Rainfall Pattern detection is one of the most important real-life problems, especially in the agricultural sector. Weather pattern forecasting is essential to understand and analyze factors like the right crop that needs to be grown, the best time to harvest, precautions to undertake on the field, etc. Weather plays a crucial role in the whole agricultural process. It is not that easy to predict the course of weather accurately and here is where Machine Learning and AI come to play. Highly trained models tuned with the right hyper-parameters can yield accurate results in forecasting. The project covers the detection of rainfall patterns using some machine learning models.

The data was taken from : https://www.kaggle.com/rajanand/rainfall-in-india?select=rainfall+in+india+1901-2015.csv

In this project, we are taking the rainfall data of Tamil Nadu, and based on the rainfall pattern during a three-month window, we try predicting the succesive month's rainfall pattern. Since this is a very basic project, a lot of importance wasn't given to the tuning of the hyperparameters. We have used three ML models - Random Forest, Lasso and Linear Regression models.

This project has been designed to predict the rainfall for one state in India, Tamil Nadu. Rainfall Prediction is one of the most difficult and uncertain tasks that has a significant impact on human society. Accurate and timely rainfall prediction can proactively help reduce human and financial loss. This project was started with the hopes that it will help the farmers and other people in the agricultural industry to choose their crops wisely for the harvest season so that they would not have to face any loss.

As of now the model has been trained for the above mentioned state.It can further be expanded to the other states of India. The program will also be made general by making it pick the best model using cross validation and accuracy score to predict the rainfall. The accuracy of the model can also be boosted using methods like algorithm tuning, feature selection etc. Other machine learning algorithms can also be used to predict the results.

https://github.com/vgaurav3011/Rainfall-Prediction/blob/master/Exploration_Rainfall_Data.ipynb