Designing a data analysis to remove risk

I'm getting ready for Spring in the garden but starting too early risks losing some plants to sudden cold.

Last week I found a set of data with Portland's high and low temperatures from 1938 to today in a simple format.

So I loaded that data up, ran a simple analysis, and found how many days in March dipped below certain temperature (freezing and 25 F). It also showed me the last years that happened and how frequent they were.

Using that data I could see what the risks are to plant some specific cover crops. Since the seed was inexpensive and the risks were low, I decided to go ahead and plant the first round early and see if it can take advantage of an extra month of growth.

Best case, I get a solid month of growth which will help out the soil immensely (and keep dogs and people from walking on the beds). Worst case, everything gets wiped out and I'm out 11 cents of seed.

Given how fun I find data analysis, I'll probably play with that temperature data even more and see what other insights I can pull from it. Maybe this year I'll even get my own weather station and analyze our backyard data.

Data and analysis should lead to decisions. The temperature data helped me decide if I should risk planting early or wait a few weeks. Data and analysis without a decision isn't worth much.

That's why I built the Guidance section into Repeat Customer Insights. It takes common decisions and tasks you have in your store ("I want to find my best customers and how much they are ordering and spending") and explains how to use the different analyses to accomplish that.

Think of it as a supercharged FAQ for your customer, order, and product data.

Eric Davis