- Define the problem you want to solve by defining the input variables and the output variable. (Ex: Prediction or Classification)
- Explore the dataset of the underlying problem. (Ex: Probability distribution of the data)
- Divide the dataset between Train/Test/Validation.
- Choose a mathematical model or algorithm that will help you model the problem and predict the output. (Ex: Neural Networks, SVM, Linear Regression etc)
- Now, train that algorithm on your train dataset and perform prediction using Test data set. Find the performance of the algorithm using various metrics. (Ex: MSE, Accuracy, Precision)
- Follow these examples for detailed implementation of the above mentioned workflow:
- Keyword spotting using LSTMs
- Handwriting Recognition using Classification trees
- Linear Regression
How to apply the heuristic algorithms for prediction/modeling process?.
7 views (last 30 days)
Puru Kathuria on 26 Feb 2021
A general workflow is described in the below steps:
Hope it helps!