Forecasting Weather Using Machine Learning

Forecasting Weather Using Multinomial Logistic Regression, Decision Tree, Naïve Bayes Multinomial, and Support Vector Machine

---

Data set

Our dataset looks like below which we collected from Bangladesh Meteorological Department

We had last 30 years [1988-2017] of weather data.The training and test set is divided into two segments having 70% and 30% data split across the two categories.

---

Parameters:

1. Day
2. Month
3. Year
4. Humidity(%)
5. Max Temp(in ⁰C)
6. Min Temp(in ⁰C)
7. Rainfall (in mm)
8. Sea Level Pressure (in mb)
9. Sunshine (hours)
10. Wind Speed(knot)
11. Cloud (in okta)

---

Train and Test accuracy of the models tested:

Model	Training Accuracy (%)	Testing Accuracy (%)
Logistic Regression	74.2	76.9
Decision Tree	76.8	74.05
Multinomial NB	54.27	54.34
SVM	76.42	77.52