Intro to Predictive Analytics
Part of the Data Analyst Roadmap
In short
Dashboards tell you what happened. Sooner or later someone asks 'what will happen next month?' Predictive analytics uses past patterns to make a careful, honest guess about the future.
Using historical data to estimate future outcomes — a step beyond describing what already happened, and the boundary where analyst work starts to overlap with data science.
Think of it like this
Clothes on the line
You see dark clouds, the wind picks up, and it's July. You bring the clothes in before a drop falls. You're not certain it'll rain, but past monsoons taught you the pattern. Predictive analytics is the same: patterns from history, turned into a likely outcome.
Watch it happen
From history to a prediction
1/5Collect past examples with known outcomes, like two years of customers and whether they stopped ordering.
The big ideas
What you need to know
Start with a decision
Prediction is useful only if someone will act on it, like calling customers likely to leave.
Outputs are uncertain
Give ranges or probabilities. '70% likely to churn' is more honest than 'will churn'.
Test on unseen data
A model that only fits the past perfectly often fails on the future. Always hold some data back.
The past must resemble the future
If the business changed, say a new pricing plan, old patterns may not hold.
Save yourself the pain
Common mistakes
Oops
Building a model before agreeing what decision it supports.
Do this instead
Ask who will use the prediction and what they'll do differently. Then build.
Oops
Reporting a single number as if it's certain.
Do this instead
Show a range or probability and how accurate the model was on test data.
Your mission
Spot predictive questions
Quick check
Why do we test a predictive model on data it wasn't trained on?
Go deeper
- 1Read a beginner-friendly overview of what predictive analytics can (and can't) do.
- 2Identify a prediction problem in a domain you're interested in.
- 3Learn to be skeptical of predictions presented without an accuracy or error estimate.
More topics at this stop
Linear Regression for Analysts
Logistic Regression Basics
Time Series Analysis Basics
Clustering Basics (K-Means Overview)
Intro to Machine Learning for Analysts
When to Escalate to a Data Scientist
Forecasting Fundamentals
Stuck on this topic? Ask an Insider
Get 1:1 guidance from people who've walked this exact path — on the InsideEdge app.
