My first Machine Learning experiment with real time data



Solving the most common problem predicting house prices in any suburb using supervised learning.

What's amazing is we shall build a Machine Learning model using MLJAR

MLJAR is a human-first platform for machine learning.
It provides a service for prototyping, development and deploying pattern recognition algorithms.
It makes algorithm search and tuning painless!

The basics of ML can be read anywhere on github/google.
What we need to know is regression/classification and when to use what. 


First things first test data:

train sample - (data_train.csv file) for model learning 
test samples (data_test.csv) for predictions



  1. Create a new project
  2. Select Regression as a task
  3. Add train and test data
    Note - When we add test dataset check the option: This dataset will be used only for predictions because we want to predict the sale price
  4. Specify columns that we use , target as "Saleprice column" in train data
  5. Dont use it for "Saleprice column" in test data
  6. ‘Accept attribute usage’ for both data sets
  7. Create a new experiment - For learning algorithm we will use Extreme Gradient Boosting (xgboost)
  8. metric to be optimized: Root Mean Square Error. (what is this?)
  9. And click Start!


To compute predictions check Predict view


Inspired from: https://medium.com/towards-data-science/regression-predict-house-price-lesson-2-5e23bec1c09d


Comments

Popular posts from this blog

Trim / Remove spaces in Xpath?

RPA - Blue Prism, OpenSpan, Automation Anywhere vs UIPath