Data Analysis with Least Squares Regression Line

This is one of the primary ways that you can analyze raw data correlation and by using the least squares model, it makes the data much cleaner and allows the user to plug in their own range of inputs in order to see what the output set looks like based on the raw input/output (x/y) values of your raw information.





Pricing
*Note you will be sent the spreadsheet once payment received.

So, I build a nice excel template that allows the user to see how all the calculations work and can be modified to be used with any two data sets that you want to put against each other.

The simple example I used was the length of time a candle burns and the weight of the candle. It's not that hard to make an inference that the longer the burn time, the less candle. However, some relationships may not be as clear and some things you may want to test to see what their level of correlation is or if it is not correlated at all.

All of those things can be gleaned by using the least squares regression math and then applying the equation to whatever set of inputs you want to visualize.

A scatter plot has also been added to the visuals so you can see what the raw data looks like by itself.

In the excel template, I did go through and explain how each calculation works and the logic (pseudocode style).

Interesting Uses - I will update these as I think of them.

  • Set inputs (x) as years/months going forward and have the output (y) the resulting month/year revenue/ebitda/etc...If correlation is strong and positive it means as years go up, revenue/ebitda/etc..goes up. If correlation is strong and negative it means as years go up, performance goes down. Closer to 1 / -1 is strongest relation and closest to 0 is least relation.

No comments:

Post a Comment