xxxxxxxxxx
To divide each of the components of a normal by the square root of the sum of their squares. Then, if the normal is thought of as a vector from the origin to the point (nx', ny', nx'), this vector has unit length:
nx' = nx/factor
ny' = ny/factor
nz' = nz/factor
xxxxxxxxxx
In relational dbms, it is used to
- Reduce redundancy / duplications
- Increase speed of transactions
- Improves integrity
- mostly used 1NF, 2NF, 3NF
1 NF:
- First normal form
- Each cell contains single value
- Each row is unique
2NF:
- Second normal form
- Database must be in 1NF
- Separata table for values that occur frequently in multiple rows for same column
- Make sure to connect the separate table with a foreign key
3NF:
- Third normal form
- Database must be in 2NF
- If a column in a table does not depend on that table's primary key,
then separate that unrelated column
- Make sure to connect the separate table with a foreign key
xxxxxxxxxx
double Normalize(double val, double valmin, double valmax, double min, double max)
{
return (((val - valmin) / (valmax - valmin)) * (max - min)) + min;
}
xxxxxxxxxx
The process of bringing values into a standard range so that data
retains intrinsic influence on models without involving bias to the model.
Helps to reduce overfitting for large values that produce large
parameters/co-efficients in models. ways of normalizations
1) feature-scaling
2) min-max
3) Z-score
xxxxxxxxxx
// potential for div by 0 error, so moderate your inputs
fn normalize(value, min, max){
return (value-min)/(max-min);
}