Key Components:

• Input: Raw data from various sources
• Process: Statistical & computational methods
• Output: Actionable insights

Example - E-commerce:

• Descriptive: Sales dropped 15% last month
• Diagnostic: Checkout abandonment increased

• Predictive: Sales will drop 20% if unchanged
• Prescriptive: Simplify checkout, add payment options

Sources:

• Databases (SQL, NoSQL)
• APIs (REST, GraphQL)
• Web scraping
• Surveys and forms
• IoT sensors

Philippine Example:

Key Considerations:

• Storage: CSV, databases, cloud
• Quality: Accuracy, completeness
• Governance: Access control, policies
• Security: Encryption, backups

Philippine Government Data:

Key Activities:

• Summary statistics
• Distribution analysis
• Correlation exploration
• Outlier detection

# Quick EDA in Python
import pandas as pd

df = pd.read_csv('ph_data.csv')
df.describe()  # Summary stats
df.info()      # Data types
df.isnull().sum()  # Missing values

import matplotlib.pyplot as plt

# Population by region
region_pop = df.groupby('region')['population_2020'].sum()
region_pop = region_pop.sort_values(ascending=False)

# Create bar chart
plt.figure(figsize=(10, 6))
plt.bar(region_pop.index[:5],
        region_pop.values[:5] / 1e6)
plt.ylabel('Population (Millions)')
plt.title('Top 5 Regions by Population')
plt.show()

Statistical Inference:

• Drawing conclusions about populations
• Hypothesis testing
• Confidence intervals

Machine Learning (Refresher):

• Supervised: Regression, Classification
• Unsupervised: Clustering

Key insight: Not all 2-point shots are equal, and most mid-range shots are bad decisions.

James Harden's step-back 3 became the signature move of the analytics era.

Foundations (Weeks 1-3)

• Week 1: Introduction & Lifecycle
• Week 2: Probability & Statistics
• Week 3: Data Wrangling

EDA & Visualization (Weeks 4-6)

• Week 4: Exploratory Data Analysis
• Week 5: Visualization Principles
• Week 6: Storytelling & Dashboards

Modeling (Weeks 7-9)

• Week 7: Regression Analytics
• Week 8: Tree-Based Methods
• Week 9: Clustering & Segmentation

Advanced (Weeks 10-12)

• Week 10: Time Series Analytics
• Week 11: Text Analytics & Ethics
• Week 12: Capstone Presentations

Data Points Analyzed:

• British "House of Cards" was popular on Netflix
• Kevin Spacey films performed well
• David Fincher films had high completion rates
• Political dramas had dedicated audiences

Result: Massive success, proved data-driven content creation works. Now standard practice for streaming.

What They Test:

• Thumbnail images (different ones for different users!)
• Row ordering on homepage
• Synopsis text and trailers
• Autoplay timing
• Skip intro button placement

Thumbnail Optimization:

Why It's Brilliant:

• Free viral marketing every December
• Makes users feel "seen" by the algorithm
• Creates social proof and FOMO
• Competitors (Apple, YouTube) now copy it

Alternative Data Used:

• Transaction frequency and amounts
• Bill payment consistency
• App usage patterns
• Network of contacts
• Top-up behavior

This is exactly what we'll build in Week 7-8!

Fraud Signals Analyzed:

• Transaction velocity (too many, too fast)
• Geographic anomalies (login from new location)
• Device fingerprinting
• Behavioral biometrics (typing patterns)
• Network analysis (connected to known fraudsters)

ML Models Used:

• Real-time scoring (< 100ms decision)
• Anomaly detection for unusual patterns
• Graph analytics for fraud rings
• Adaptive models that learn from new fraud patterns

Factors in the Model:

• Real-time demand (ride requests)
• Available drivers in area
• Weather conditions (rain = high demand)
• Events (concerts, games)
• Time of day patterns
• Historical data for the area

Key Findings:

Translation: Black defendants who didn't re-offend were almost twice as likely to be labeled high-risk.

Potential Issues:

• GScore: Alternative credit can perpetuate inequality
• Facial recognition: Used by malls, may misidentify darker skin
• Surge pricing: Fair during typhoons?
• Content moderation: Filipino-language hate speech harder to detect

Skills in Demand:

• Python/R programming
• SQL and databases
• Data visualization
• Statistical analysis
• Communication skills
• Business acumen

# Lab 1 Preview
import pandas as pd
import matplotlib.pyplot as plt

# Load Philippine GDP data
df = pd.read_csv('ph_gdp.csv')

# Quick exploration
print(df.shape)
print(df.describe())

# First visualization
df.plot(x='year', y='gdp')
plt.title('Philippine GDP Growth')
plt.show()