Machine learning
for calculating tariff prices
Why a techtalk?
by Andrew Ng
5 courses
~80 hours of videos + coding
30 euro / month
The problem
?
?
?
?
?
The problem
3rd party service
- avg 5 seconds
- in 20% cases > 7 seconds
- sometimes > 60 seconds
The solution
- Collect the data
- Take a single tariff
- Use it to train the ai to tell:
- Is it in the list (yes/no) ?
- What is net price (decimal)
- Hope the ai solves the problem
3 months, 331726 entries
ANN - Artificial neural network
- Ramón y Cajal, Santiago (1899).
Comparative study of the sensory areas of the human cortex.
- Frank Rosenblatt build Mark I Perceptron in 1960
ANN - Artificial neural network
ANN - Forward propagation
Supervised learning,
error backward propagation
A sample training example
{
"input": {
"familyStatus": "Keine Angabe",
"educationType": "Berufsausbildung",
"jobSituation": "Angestellt/Selbstständig",
"industry": "Sonstige Branche",
"benefitAgeLimit": 60,
"benefitAmount": 1000,
"fractionOfficeWork": 100,
"staffResponsibility": 0,
"smoker": false,
"birthday": "1983-11-07",
"insuranceStart": "2018-12-01"
},
"output": {
"exists": true,
"net": 92.05
}
}
Enum, 6500 options, actually ~ 6 risk groups
Enum, 4 options
Enum, 9 options
Enum, 3 options
Enum, 10 options
Number
Number
Number
Number
Boolean
Date
Date
Boolean
Number
"occupation": "Isolierer,in",
The input layer
Input layer
single node,
a real number 0..1
(the presence problem)
41 input nodes
Hidden
layer
Output
layer
The code itself is here:
https://github.com/pavelgit/ml-techtalk
or here:
https://bitbucket.check24.de/users/pavel.liubinski/repos/ml-techtalk-2019/browse
This was easy!
- Is it a professional approach? - how to tune, where to run, how long to wait?
- Single occupation - try a bottleneck layer?
- Max error still huge - try another loss? another learning algorithm?
- "Predictability" - will it always return "sane" results?
Questions remaining
Machine learning for calculating tariff prices
By Pavel L
