Data Scince

Data Science

Two types of ml algorithm training.

• Prediction:
  Unsupervised learning (unknown data) (nearest neighbours)
• Exploration:
  Supervised learning (known data)

---

Data Mining

• Models

• clustering

• Business understanding-> data Understanding-> data Prepare->modeling(used be done manually with large scales mathematic analysis)(Machine Learning)>evaluation->deployment

• Machine Learning:
  Getting computers to act without being explicitly programmed.
  Learning from data.

• Statistical learning.

---

Disadvantage:

• untrustworthy resource
• Spark ML/Spark MLLib
• bring potential suitable modelling lists

---

Case Analysis:

Sad Francisco Or New York
Some intuition(elevation)

Adding nuance
(Adding another dimension allows more nuance)
[scatterplot]
(Features, predictors, variables)

Drawing Boundaries
Scatterplot matrix

Decision tree(one variable each time)
Find a Better Boundaries[histogram]
Forks
Split point
Over fitting