This is first in the series of notes that I plan to put online which I take during my online courses. This one specifically is from the Udacity Course Classification Models.

The course start with a discussion around a problem where we need to predict which things people are more likely to buy with the change of weather.

Lesson 1:

• Discussion about examples of classification problems. For eg. determining if a soyabean has a disease or not using an image.
• Discussion about binary(either of the two values) and non-binary(categorical variables) examples.
• Discussion about Statistical Terms - Target and Predictor Variable.
• Target Variable - Field we are trying to understand and predict - Dependent Variable.
• Predictor Variable - Used to predict the target variable - Independent Variable.
• Remove Duplicate Variables(one variable subset of other) identifying correlations.
• Correlation - Measure of association between two variables whose vaues lie between -1 to 1.
• Three types of associations to be studied for understanding correlation
  • Pearson correlation
  • Spearman’s Rank correlation
  • Hoeffding’s Independent Test
• Pearson Correlation - Correlation Plots
  • No issues during training but issues while testing or predicting.

Key take outs:

• For those who are already aware of Machine Learning, Predictor Variable is what we call as a feature and Target Variable is the objective function.