MachineLearning

Day 1

• Pandas module
  1. Convert dictionaries to Dataframes
  2. Slicing dataframes
  3. Making new columsn in dataframes
• SKLearn and Quandl module
  1. Get financial and economic datasets using Quandl
  2. Performing mathematical operations on dataframe columns
  3. Dataframe functions - .head() .tail() .shift() .fillna() dropna()

Day 2

• Train, test, predict data using Linear regression or Simple vector machine model
  1. Features vs labels
  2. Training and predicting using a model
     1. Prepare training data and split in 2 parts, ~80% to train ~20% to test [ model_selection.train_test_split() ]
     2. Define a classifier/model, like LinearRegression, SVM (Simple vector Machine) and then Train the classifier using .fit()
     3. Test accuracy of the classifier with respect to test data from step 1 [~20% of data]
     4. Predict - Label = classifier.predict('Features')
• Best fit line and how regression works
  1. What is slope(m) and intercept(b)
  2. Linear Regression = mX + b

Day 3

• What are Squared error?
• Squared error vs Absolute errors
• R-Squared / Coeffcient of determination
• Classification with K-Nearest neighbor (KNN)

Day 4

• Euclidean distance

  

• Making your own k-NN (k-Nearest Neighbor) algorithm in python

• Comparing the accuracy and confidence of your algorithm with SKLearn module's neighbors.KNeighborsClassifier()

• Accuracy vs confidence in k-NN algorithm

Day 5

• SKLearn Support Vector Machine (SVM) classifier
• Making your own Support Vector Machine (SVM) algorithm in python [Courtesy: ]