Demand Planning Class Notes
Nicole-line breaks mean new slide
important questions
Forecasts are needed to predict demand all different teams within the company need the forecast different users have different time requirements and detail reqts you might have to collect more data if you don't have enough cost depends on the scope of the project need to engage the users, so have to provide a feedback system
The top chart appears to be a ore difficult to forecast but they just narrowed the y axiz
2nd chart down slope is almost random Us treasure bills
Sales of product c chart is probably affected by seasonality
Subjective judgmental approach relies on your personal experience
Very biased because it is based on your personal experience and also bc humans have a strong reaction to recent events aggregate-looking at something from a larger picture. Ie not the I phone but all apple products, or not this month but the whole yaet
-salespeople bias bc they will predict less sales so that they can meet and exceed the #
Relational approach-you believe there is a reason for things happening, so you look for a causal relationship between demand and the generative factor
-factors drive demand
-regression technique -regress causal relationships to several causal factors to I'd relationships or correlation
-Downside-need real world data which may be hard to find hard to find the relationship in real world data
experimental approach
time series approach
-look for patterns in data
-pattern vs noise, and noise is random and has zero average ideally
-in real world. Noise is random but probably not with zero mean
-time series usually decomposed into different affects (seasonality, tend, noise, etc) d(t) =(L+
back fit your data to see if you have a good forecast
what would you do
1. Plot the data in excel forecast approaches were in order with charts
moving average-look at last few periods and update each new period with the new data (ie in
important questions
Forecasts are needed to predict demand all different teams within the company need the forecast different users have different time requirements and detail reqts you might have to collect more data if you don't have enough cost depends on the scope of the project need to engage the users, so have to provide a feedback system
The top chart appears to be a ore difficult to forecast but they just narrowed the y axiz
2nd chart down slope is almost random Us treasure bills
Sales of product c chart is probably affected by seasonality
Subjective judgmental approach relies on your personal experience
Very biased because it is based on your personal experience and also bc humans have a strong reaction to recent events aggregate-looking at something from a larger picture. Ie not the I phone but all apple products, or not this month but the whole yaet
-salespeople bias bc they will predict less sales so that they can meet and exceed the #
Relational approach-you believe there is a reason for things happening, so you look for a causal relationship between demand and the generative factor
-factors drive demand
-regression technique -regress causal relationships to several causal factors to I'd relationships or correlation
-Downside-need real world data which may be hard to find hard to find the relationship in real world data
experimental approach
time series approach
-look for patterns in data
-pattern vs noise, and noise is random and has zero average ideally
-in real world. Noise is random but probably not with zero mean
-time series usually decomposed into different affects (seasonality, tend, noise, etc) d(t) =(L+
back fit your data to see if you have a good forecast
what would you do
1. Plot the data in excel forecast approaches were in order with charts
moving average-look at last few periods and update each new period with the new data (ie in