![]() To get started, we can predict the value for just one datpoint, known as a point prediction. ![]() The regression can then be used to develop forecasts for the future, which is an example of predictive analytics.Īs the article implies, however, it’s simple enough to turn a regression into predictive analytics - just start making predictions! Let’s try that now. When regression analysis is used to explain the relationships between variables in a historical context, that’s an example of diagnostic analytics. Linear regression can be done pretty quickly in Excel using the Data Analysis tool. Regression can give us the big picture about relationships between variables. That makes basic regression a form of diagnostic analytics, or analysis about why things happened in the past.Īs this article from Harvard Business School puts it: Regression analysis is in almost every type of statistical software like SPSS, R, and not to mention Excel. This tells us about patterns in past data, but doesn’t in itself make predictions about the future. Think about how regression works: we fit a line to our data and decide how well that line describes a relationship between independent and dependent variables. To make matters more complicated, we didn’t really just do predictive analytics anyway with this regression model, although we could have… Linear regression and predictive analytics That puts predictive analytics near the AI camp, but not in it. Making predictions is a typical human task that can be done “artificially”, but few would expect humans to make predictions the way that computers do, by following a strict statistical model. Predictive analytics is using data and statistical models to forecast what will happen in the future. AI and predictive analyticsĪrtificial intelligence is what it sounds like - tasks done by computers that normally would require human intelligence, such as transcribing speech, detecting pictures and so forth. So does linear regression in XLMiner count as AI? Sort of - it depends on how the regression model is used. These days, it can seem like a data product is a snoozefest unless it’s AI-powered. Aside from these measures, what kind of analytics are we doing? Is this AI? Does this count as AI? If you’d like to learn more about interpreting these, check out Advancing into Analytics. Here you have typical regression diagnostics such as coefficient p-values, R-square and more. You should see the following output from XLMiner after running the regression: XLMiner can also provide additional outputs to help in checking these assumptions. To be fair, Python and R provide much more robust environments for these regression checks, but if you’d like that hands-on approach that only Excel can provide, check out my book Advancing into Analytics: From Excel to Python and R. It can be tempting to jump right into building models and making predictions as we did, but in practice it’s necessary to explore the data and check whether it meets the assumptions of whatever model you’re using. Unfortunately, it can be difficult to use the drag-and-drop feature in XLMiner to name an input range, so you may need to physically type in the cell locations.
0 Comments
Leave a Reply. |