In three videos we’ll introduce JMP’s Predictive Analysis Process for New Users.
Goal: Demonstrate how we can use JMP to predict potential outcomes using both single and multiple sets of variables or influencers.
Story Setup: Spin Manufacturing creates plastic toys and they're asked to predict how many toys can they produce in a single year with their two existing assembly machines. The variables include daily-recorded inside facility temperatures, an old machine or a new machine, employee level (years on job), high or low facility lighting, and daily production run quantities.
Part 1 - Video 1
We're going to cover JMP's Predictive Analysis Process and we’ll start with setting up our data.
- Data Table Prep, Column sorting, Recode, Repeating Sequences, Sorting.
- Saving Script (Analysis) to Data Table
- Bivariate Modeling
- One way regression analysis (predicting Y with one single X variable (Factor)
- Using the Crosshairs tool to predict values in a regression
- Adjusting report settings in the Preferences panel
- Fitting a Line
Part 2 - Video 2
We are now going to take our same data table and predict outcomes based on multiple variables.
- Multivariate Analysis using Fit Model
- Display Settings>Mean Diamonds
- Display Settings>Point Jittered
- Residual by Predicted Plot report.
- Adjusting Linear Fit - Set Level > .05
- Fit Mean
- Signal/Noise
- Linear Fit> Summary of Fit> R Values
Looking at the Summary reports below we can see the…
The Plot Residual By Predicted, which shows a shows the residuals plotted against the predicted values of Y. You typically want to see the residual values scattered randomly about zero.
Part 3 - Video 3
We’ll now use JMP’s profiler to predict outcomes by interactively altering our variables.
