09-03-2020 00:22 AM
Every day we are inundated with noise about the health of the economy. Take for instance the following headline, “Consumer Sentiment ticks down 2 points to 93.1”. Without more information, there is no way to know if this is slightly negative, very negative, or simply neutral.
The goal of this exercise was to use the advanced capabilities of Power BI to:
From a visual analytics perspective studying simple averages is a great place to start. Couple this with another powerful methodology called Decile Analysis (also called segmentation or binning data into equal 10-part N-Tiles) and you are on a path towards an advanced understanding of your data.
This dashboard uses this approach to study 8 leading indicators (grouped into equal 8-part tiles based upon how much the indicator has changed over the prior 3 months) to show the average future change in the employment rate, housing prices, or broad stock market index.
Selecting the Data
The Federal Reserve site FRED hosts a wealth of economic data available at https://fred.stlouisfed.org/. To select data elements, I first studied the Conference Board’s Index of Leading Indicators and searched for those data elements on FRED.
Some issues I encountered included:
Here are some advanced capabilities of Power BI that I utilized:
A great dashboard leads to interesting discoveries within the data. While many of these statements are things I have heard before it was great to prove them to myself.
What interesting discoveries can you make?
Thank you for this great dashboard. It takes time to learn it to read. Could you please comment my findings?
1. On the first graph there are the "Avg 10-12 mo change" value and the "Current value" on top of it... I noticed that there's a shift on Date axis, befause the future value graph is 12 months shorter (see attachement Clipboard01.jpg). Is it intentional or is it a mistake?
2. Am I right, that the Historical N-Tile Analysis is to show the quality of the model? When there are batterns on the rows, then the model results can be trusted?
3. Is the "Current N-Tile" meant to show if we are currently out of the average?
4. On the bubble graph, what does the bubble size represent?
1) Thank you for your feedback. I have made some updates to this visual to make this more clear. There should be a 12 month lag between when the indicator (e.g. Consumer Sentiment) reading occured, and when the actaul change was measureable.
2) First note this is an informational dashboard vs a mathmatical model so I'm hesitant to use the word "trust". When there are stronger patterns on rows the indicator has has more consistent past results.
3) The bottom right visual is meant to show what has happened historically over the future 12 months for each of the indicators. This shows the current reading for each indicator.
4) The bubble size is the same as the color, larger means a larger historical average future move.