You are on page 1of 3

REPORT

19 FEB – 25 FEB

Explore Machine Learning and Regression Learner

Machine Learning

Explore about the fundamental of machine learning. Machine learning can be divided into two,
Unsupervised learning and Supervised learning.

Unsupervised Learning finds for patterns in datasets that does not have labeled response and helps to reduce
variables in data. It is good method to segment data. Technique used for this method is Cluster analysis which
separates data into groups based on same characteristics. Cluster analysis consist of several algorithms:

1) K-means, K-medois, Fuzzy C-means


2) Hierarchical
3) Gaussian Mixture
4) Neural Network
5) Hidden Markov Model

Supervised Learning needs correct input and output dataset to develop trained model that can be used to
forecast future dataset. The predictive dataset can be trained by using Classification or Regression techniques.
Classification techniques predict any discrete response and classified the predictive model into different groups
while regression techniques predict continuous response. Regression techniques are suitable for predictive output
that can accept range of values.

Regression Learner

Regression learner app is one of supervised machine that can be used to trained regression models to forecast
data. This app develops multiple regression model such as linear regression, regression trees, support vector machine
and gaussian process regression models. The best model is choose based on the validation errors such as RMSE,
MSE, MAE and R-squared. The best fit model will have the smallest value of RMSE.

In this assessment, two different datasets have been used to develop predictive model for forecasting
purposed. The first dataset is to predict Commercial, Public Lighting, Domestic, Industrial and total sale data. Each
model is trained separately.

Figure 1 : Compares Original data and Predicted Data


Figure 2 : Plotting graph of total sales

Figure 3 : Graph of predicted load

26 FEB – 28 FEB

Get familiar with basic Simulink

Get familiar with Simulink features and blocks and build simple models and simulate the system.

- Create comparisons, decision statements and subsystems models


- Create discrete and continuous systems model
Figure 4 : Comparison model

Figure 5 : Comparison model

Figure 6 : Signal generate from decision model

You might also like