Machine learning key terms basics

ML - Key terms 

-> Classification - Identify a group/category from bunch of test data with given input data
-> To Classify a input we need to train a machine with algorithm with test data - training set
-> training set have N no of feature and one target variable
-> target variable is what we are trying to predict with ML algorithm.

-> In training set data target variable is known. 
-> Machine learn to find the relationship between feature and target variable - target variable is also known as class
-> To train a ML algorithm we need training data(X axis) and test data(Y axis)
-> To find desired level of accuracy is called knowledge representation
-> Regression is prediction of a numeric value
-> Classification and Regression is supervised learning

-> Non-supervised learning -> clustering - is a task where we group similar items
-> Non-supervised learning we will not have target value on data set.
-> Non-supervised learning reduce the data from many feature into smaller number of data, so we can visualize 2D or 3D format(eg: chart)
-> we need to find statical value that described data -> destiny estimation
-> training steps is not needed here because we will not have target value to train a machine.

-> Supervised learning - trying to predict the target value
-> if your target value is Yes/No or 1/2/3 or True/False this is called classification
-> if your target value is number of values like 0.00 to 100.00 or -9 to +9 is called regression 
-> Non-supervised learning - trying to find group for given data
    -> If you are trying to find group for given data is called clustering
-> If you need some numerical estimate to how strong the fit with each group , this is called Destiney estimation algorithm

-> To make better ML application we need to spend quality of time with data set. 

-> Steps of ML algorithm
-> create data
-> prepare input data
-> analyze input data
-> train the algorithm - Non-supervised does not required this steps
-> test algorithm
-> use it


-> Knn algorithm used to classify the data..  

continue in upcoming post...