Question

Give a general description of how fitting a curve to data works. You don’t need to get into all of the details in the book, Just give a “50,000 foot” view in your own words

Answer 1

Curve fitting is find the function or model that fits the given set of data collected. The curve developed can used for data visualisation or finding the dependent value using new set of independent variable or sometime find the values where data is not available or missing.

For example in polynomial Regression, we can fit one degree, two degree or 3rd degree polynomial. Sometime we can use log, exponential or S-Curve. Using all the curve possible, we find out the best curve which justify the current data. But we should keep in mind that sometime overfilling work really well for given data but it's performance is really poor in new data that was not used in model. So it is important that you should not overfit the data and see it works fine with hidden data. Best way to do this is split the data in training data and test data. Make model using training data and test the model in test data.