GIGAMIND

Folder:
108 Data Analysis
File:
108.30.30 Data Analysis - Prescriptive analysis

Prescriptive analysis

The goal of prescriptive analysis is to provide a recommendation (prescription) on what actions to take to achieve some objective goal (1)

In order to give a good recommendation to solve a problem it needs to be well-defined. A problem has
- A set of goals we want to achieve
- A set of actions to take to achieve the goal
- A model which describes how the actions impact the goal

An abstract example:
- Goal = maximize revenue
- Action = set different prices
- Model = how does setting price change the revenue from selling the product?
- Two forces in play: When we lower price we sell more, but we also get less revenue per product sold. And vice versa.
- This is called a tradeoff << in most models identifying the tradeoff beforehand helps to understand what we are looking to solve

Revenue is Quantity * Price.

In this example, we know the regression line and can feel comfortable that we understand how much of the product will be sold at each price. So we basically just multiply the price times the quantity for each value, which produces a curved plot, and choose the price that produces the maximum revenue. Simple.

Now the model can get more complex. We can optimize for profit - which means we must consider cost to produce the item, cost to market the item, etc.
- To maximize profit we must take costs into account
- Usually the optimal price can be found by using the sales regression, combined with costs at each price point, until the gain in revenue from a higher price is equal to the increase in costs.
- Marginal Revenue (MR) = Marginal Costs (MC)
(2)

References:
1. https://www.coursera.org/learn/wharton-customer-analytics/home - Week 4 slides, Berman, slide 3
2. https://www.coursera.org/learn/wharton-customer-analytics/home - Week 4 slides, Berman, slide 23


Source:
  • Me