lmhsir Goto Github PK

2019-startcode

Initial project for the MIPT students: Generalised model selection

edx_capstone

GitHub for the course "Data Science: Capstone" by edX and HarvardX

forestfires

We are going to study a set of data used to predict the total burnt area of forest fires. According to specialists in metearology, the 4 main variables in weather for this case is wind, relative humidity, temperature and rain. Other variables in this dataset that may help in predicting the area are the coordinates (x and y), the month and the day. And finally we have 3 codes and an index. We will do a brief explanation of these 4. The 3 codes are FFMC (Fine Fuel Moisture Code), DMC (Duff Moisture Code) and DC (Drought Code). The FFMC is about ignition and spread of fire and is obtained by using all the 4 meteorological variables presented before. The DMC and the DC affects the fire intensity. DMC is obtained with rain, relative humidity and temperature and DC with rain and temperature. The ISI (Initial Spread Index) denotes the velocity spread of the fire and is calculated using the FFMC and the wind. These 4 were kept in the dataset since they were all directly affected by the 4 weather variables.Therefore the BUI (BuildUp Index) which is calculated using DMC and DC and the FWI (Fire Weather Index) calculated using ISI and BUI were omitted.Although the SVM (Support Vector Machine) is the best method to predict the area according to studies, we are going to do a multiple reggression study with this dataset to predict the area of burnt forest at first. And them, if we can't get an ideal model, we will try the SVM.

forestfires-1

forestfires-2

Practice data science

forestfires-3

Aim is to predict the burned area of forest fires

forestfires-4

forestfiresimulation

Simulating a forest fire in python.

regression

Metro Interstate Traffic Volume

tabular-data-projects

Collection of ML scripts for tabular datasets