This repo contains , the project that I worked on during Summers 2016 at Bhaskaracharya Institute For Space Applications and Geo-Informatics.
Project Objective : Prediction of Land Use and Land Cover
Tools Used : QGIS , Python3
Description : The algorithm uses basic linear regression to Predict "Vegetation , Soil , Water , BuildUp" in any given year in future , given its trained properly
Each Training set uses 2 training images's data features and their repective years.
The feature vector contains features namely
1. Vegetation (of Year1)
2. Water (of Year 1)
3. Soil ( of Yeaar 1)
4. BuildUp(of Year 1)
5. TimeGap between 2 images who's data has been provide (i.e Year2-Year1)
Results vector(of traing set) will include :
1. Vegetation (of Year2)
2. Water (of Year 2)
3. Soil ( of Yeaar 2)
4. BuildUp(of Year 2)

PreProcessing : this is where you'll have to use QGIS for semi automatic classification to get the features required above and put them into files.
You can read about the same in a short tutorial given in the link given below
http://semiautomaticclassificationmanual-v4.readthedocs.io/en/latest/Tutorials.html#data