Please enable JavaScript.
Coggle requires JavaScript to display documents.
(((Modeling (Causal Modeling (what events influence?, what actions…
Modeling
Classification
Individual
Identify Classes
Scoring Model allows probability of this
Mutually Exclusive(usually)
Yes or No? Will they do it?
Regressions
Individual
Numerical value of some variable for them can be found
How much will they?
Similarity Matching
Similar individuals identified
Use historical data
Clustering
Group individuals
Similarities
Not driven by purpose
Co-occurrence grouping
Association rule discovery
Market Based Analysis
Frequent itemset ming
Entity to Entity
Associations
Transaction based
Recommendations on actions
Profiling
Individual
Characterization
Compared to a population
Link Prediction
Predict Correlations
Estimate strength of links
Form basis of recommendation
Search for unexisting connections
Data Reduction
Causal Modeling
what events influence?
what actions influence?
can include randomized control experiments
experimental and observational
counterfactual analysis
Supervised vs Unsupervised
Do you have a specific target?
If yes, supervised
specific purpose
Classification
Regression
Causal Modeling
if no? unsupervised
no specific purpose or target defined
clustering
co-occurence
grouping
profiling
either
similarity matching
link prediction
data reduction
Data Reduction
Large data to small data
Can make data easier to deal with
Keep crucial information
Data Preparation
Idiosyncratic?
Define crucial variables
Conversion
Correcting Variables
Consider a leak
Will the information only exist after event?
Evaluation
Deployment after its evaluated
Uses
Find patterns and build models
using results of data mining
Process
Reasonable consistency
Repeatability
Objectiveness
