Machine Learning
Phase of an Machine Learning model build;
- Gather data from a data source (e.g databases,csv file)
- Apply EDA on the data and preprocess if needed .
- Create an ML model according to needs and problem.
- Feed data into model.
- Turne the hyperparamiters in need.
ML Terminology
- Future:is the information that used for prediction by our model.
- Predictied Value: prediction made by our model.
- Target Or Label: Real value of an example
How ML work?
DATA — — -MODEL — — PREDICT — — A)Succesful Prediction B) Failed Prediction
WHERE CAN YOU USE ML?
- Detecting spam e-mails
- Personalized ADS
- Disease detection & Diagnose
- Voice Asistans
- Autonomous Vehicles
ML PROBLEM TYPES
- Regression;
Simple Linear Regression: y=b0 + b1*x1
y=Predicted value
X1=Features
Multiple Linear Regression:y=bo +b1.x1+b2.x2+……..bx.xn
y=dependent value(target variable)
x1,x2….Xn =İndependent variable,features
REAL LIFE EXAMPLE:Predict hause Price value
y=wx +b
y=label,target(price)
w=weight of x
x=feature(squarmeter)
b=bias
Non-Linear Polynomial Regression:
y=a.(x**3) +b
Evaluation of A Linear Regression Model:
Mean Squared Error (MSE)
Mean Absolute Error: (MAE):
Underfitting / Overfitting & Good Fit,robust
Loss Reduction Method (train-validation split)
Early Stopping:
Gradient Descent Algorithm
Repeat Until Convergence:
Classification:
Logistics Regression:
Conflusion Matrix:
Accuracy,precision,recall
Classification Threshold:
ROC Curve:
Splitting Gini:
Decision Trees: