How to predict correlation between variables of the provided dataset.

[Luuv01]

Hi.

Please could anyone help with my task below:

How can I use the regression or artificial neural network methods to build up a model that may
seek to establish relationships between the variables of the provided dataset. For instance, I want to know if there are correlations between some category of products in the given dataset to answer the question do
people who buy products from the baked goods section also buy products from the Cheese section
etc? (the unit sold and sales price and other variables are given as shown below)

Thanks

[atcodedog05]

Hi @Luuv01 

This is more of a market basket analysis. Below resources might help you.

https://community.alteryx.com/t5/Data-Science/Market-Basket-Analysis-102-Alteryx-Designer-Python/ba-...

https://community.alteryx.com/t5/Data-Science/Market-Basket-Analysis-101-An-Introduction/ba-p/661963

Hope this helps 🙂

[AngelosPachis]

Hi @Luuv01,

The Association Analysis tool allows you to find if there is a correlation between a target variable and a predictor variable, and it does that through either the Spearman, Pearson or Hoeffding correlations. However, those correlations will mainly investigate linearity, so if the units sold has a linear correlation to the sales price.

From your final question, it doesn't seem that you are looking to find correlation, but rather you are interested in Market Basket Analysis. As you said, do people who buy X product also buy Y products. Those tools fall under the Predictive grouping tool palette.

Here are two posts that will help you get started with it

https://community.alteryx.com/t5/Data-Science/Market-Basket-Analysis-101-An-Introduction/ba-p/661963

https://community.alteryx.com/t5/Data-Science/Market-Basket-Analysis-102-Alteryx-Designer-Python/ba-...

Cheers,

Angelos

[Luuv01]

Thanks @atcodedog05

I will check it out

[Luuv01]

Thanks AngelosPachis,

I will check it out

[Luuv01]

Hi  @AngelosPachis 

I tried MB but it seems not working in my case, my dataset only gives the Unit sold for each product which are grouped in section so we need to build a machine learning model to predict the correlation of the section to answer the question. Do you know how we can build either Regression or Neural Network model to predict if there are any correlation between sections. Thanks

[AngelosPachis]

Hi @Luuv01 ,

Can you please elaborate more on what you want to build? From what you said earlier it seems you want to predict the number of units Sold (target variable) against the section (predictor variable).

Is that correct?

[Luuv01]

Hi @AngelosPachis

Thank you so much for you reply.

I dont wanna to predict the number of Unit Sold agaist the Section, they are already given as snipped picture of dataset below:

In stead, I want to build a model that can predict the relationship between the Section to see if people who buy products from the baked goods section also buy products from the Cheese section etc. So from the dataset above could we build that model?  Thanks

[apathetichell]

First off - change all of your columns names. R won't take in column headers with spaces in the title. Second - try the samples which are provided with Alteryx through the help tool to see how to implement Market Basket and Association Analysis - the specifics of how to get it to work may be complicated the first few times you run it - so try to adjust your data to match the workflow.

Since you are looking for the correlations - you'll want to read the report - the model won't be as helpful for you - so make sure you have a browse/render with the report anchor form your tool.

Also - reading what you want - you may want multiple logistic regressions - to test each posbility. note the specific question you are asking seems to be binary (ie. did they or didn't they also shop in the cheese section). You can convert this to 1 (they did shop) and 0 (they didn't shop) for a logistic regression model. You can do the same for the other variables if you are looking to build out models.