![Min Max
Normalization
This technique re-scales a feature or observation value with
distribution value between 0 and 1 or a given range.
Min Max shrinks the data within the range of -1 to 1 if there are
negative values, and can set the range like [0,1] or [0,5] or [-1,1].
This technique responds well if the standard deviation is small
and when a distribution is not Gaussian.
sklearn.preprocessing.MinMaxScaler](https://image.slidesharecdn.com/featurescaling-200930124149/85/Feature-scaling-9-320.jpg)
![Unit Vector Scaling
THIS SCALING TECHNIQUE IS DONE
CONSIDERING THE WHOLE FEATURE VECTOR
BE OF UNIT LENGTH.
UNIT VECTOR SCALING MEANS DIVIDING EACH
COMPONENT BY THE EUCLIDEAN LENGTH OF
VECTOR (L2 NORM).
UNIT VECTOR TECHNIQUE PRODUCES VALUES
RANGE [0,1]. WHEN DEALING WITH FEATURES
WITH HARD BOUNDARIES, THIS IS QUITE
EX. WHEN DEALING WITH IMAGE DATA, THE
COLORS CAN RANGE FROM ONLY 0 TO 255.](https://image.slidesharecdn.com/featurescaling-200930124149/85/Feature-scaling-15-320.jpg)
Feature scaling
- 2. What is Feature Scaling? Feature Scaling is a to standardize the independent features in the data in a fixed range. handle highly varying magnitudes or values or units.
- 3. Why feature scaling(Standardization)? It’s a step of Data Pre- Processing which is applied to independent variables or features of data. It basically helps to normalize the data within a particular range. Sometimes, it also helps in speeding up the calculations in an algorithm
- 4. When to do scaling? When we use below algorithms feature scaling matters: K-nearest neighbors (KNN) K-Means Principal Component Analysis(PCA) gradient descent When we use below algorithms feature scaling doesn’t require: Algorithms those rely on rules CART Random Forests Gradient Boosted Decision Trees
- 5. Distance calculation using different technique: Euclidean Distance : It is the square-root of the sum of squares of differences between the coordinate , X is Data Point value, Y is Centroid value and K is no. of feature values.
- 6. Continue.. Manhattan Distance : It is calculated as the sum of absolute differences between the coordinates (feature values) of data point and centroid of each class.
- 7. Continue.. Minkowski Distance : It is a generalization of above two methods.
- 8. Feature Scaling Techniques Min-Max Normalization Standardization Max Abs Scaling Robust Scaling Quantile Transformer Scaling Power Transformer Scaling Unit Vector Scaling
- 9. Min Max Normalization This technique re-scales a feature or observation value with distribution value between 0 and 1 or a given range. Min Max shrinks the data within the range of -1 to 1 if there are negative values, and can set the range like [0,1] or [0,5] or [-1,1]. This technique responds well if the standard deviation is small and when a distribution is not Gaussian. sklearn.preprocessing.MinMaxScaler
- 10. Standardization This technique is used to re-scales a feature value so that it has distribution with 0 mean value and variance equals to 1. scaling happen independently on each feature by computing the relevant statistics on the samples in the training set. If data is not normally distributed, this is not the best Scaler to use. sklearn.preprocessing.StandardScaler
- 11. Max Abs Scaling Scale each feature by its absolute value. This technique scale and each feature individually such that the maximal absolute value of each feature in the training set is 1.0 and minimum absolute is 0.0. On positive-only data, this Scaler behaves similarly to Min Max Scaler. sklearn.preprocessing.MaxAbsScaler
- 12. Robust Scaling This Scaling technique is robust to outliers, If our data contains many outliers, scaling using the and standard deviation of the data won’t work well. This Scaling technique removes the median and scales the data to the quantile range(defaults to IQR: Interquartile Range). sklearn.preprocessing.robust_scale
- 13. Quantile Transformer Scaling This technique transforms the features to follow a uniform or a normal distribution. A quantile transform will map a variable’s probability distribution to another probability distribution. transformation is applied on each feature independently. First an estimate of the cumulative distribution function of a feature is used to map the original values to a uniform distribution. The obtained values are then mapped to the desired output distribution using the associated quantile function. Then a Quantile Transformer is used to map the data distribution Gaussian and standardize the result, centering the values on the mean value of 0 and a standard deviation of 1.0. sklearn.preprocessing.quantile_transform
- 14. Power Transformer Scaling The power transformer is a family of parametric, monotonic transformations that are applied to make data more Gaussian-like. This is useful for modeling issues related to the variability of a variable that is unequal across the range. sklearn.preprocessing.power_transform The power transform finds the optimal scaling factor in stabilizing variance and minimizing skewness through maximum likelihood estimation.
- 15. Unit Vector Scaling THIS SCALING TECHNIQUE IS DONE CONSIDERING THE WHOLE FEATURE VECTOR BE OF UNIT LENGTH. UNIT VECTOR SCALING MEANS DIVIDING EACH COMPONENT BY THE EUCLIDEAN LENGTH OF VECTOR (L2 NORM). UNIT VECTOR TECHNIQUE PRODUCES VALUES RANGE [0,1]. WHEN DEALING WITH FEATURES WITH HARD BOUNDARIES, THIS IS QUITE EX. WHEN DEALING WITH IMAGE DATA, THE COLORS CAN RANGE FROM ONLY 0 TO 255.
- 17. Thank You