Ch.01 Mini Project 02 — Dataset Type Explorer

Concept: Ch.1 data structure taxonomy — Record, Data Matrix, Document, Transaction, Graph, and Ordered data. Each has a characteristic shape that can be detected from column names and value patterns.

What it does

Given any CSV file, dataset_type_explorer.py will:

1. Classify the dataset into one of the six Ch.1 structure types
2. Report confidence level (High / Medium) and the evidence behind the decision
3. Show a column-type breakdown (numeric vs. text columns)
4. Print a structure preview — first few rows in aligned columns
5. Link the result back to Ch.1 with a short reference note

Detection Logic (priority order)

Priority	Signal	Detected Type
1	Column name contains graph keywords (source, target, node, edge)	Graph Data
2	Column name contains time/sequence keywords (date, timestamp, year)	Ordered Data
3	≥70% of columns are binary (0/1) and ≥5 columns	Transaction Data
4	≥70% of columns have word-like names AND numeric values	Document Data
5	All columns are numeric	Data Matrix
6	(fallback) mixed types, no structure detected	Record Data

Usage

# Run the built-in demo (four types shown back-to-back)
python dataset_type_explorer.py

# Classify your own CSV
python dataset_type_explorer.py my_dataset.csv

No external dependencies — pure Python 3 standard library only.

Sample Output

================================================================================
  DATASET TYPE EXPLORER
  Source  : Demo: Transaction Data (binary item matrix)
  Rows    : 5   |   Columns: 5
================================================================================

  Detected Type : Transaction Data
  Confidence    : High
  Shape         : 5 transactions, 5 items (columns).

  Evidence:
    - 5/5 columns are binary (0/1) - looks like item-presence flags.
    - Wide binary layout is characteristic of transaction/basket data.

  Column breakdown:
    Numeric  ##### 5
    Text      0
--------------------------------------------------------------------------------
  Structure Preview:

  milk            bread           butter          eggs            juice
  ...

Key Concepts from Ch.1 Applied

Concept	Where it appears
Six data structure types	classify_dataset() return value
Binary / numeric detection	_is_binary_col(), _is_numeric_col()
Graph keyword matching	GRAPH_KEYWORDS set
Time/order detection	ORDERED_KEYWORDS set
Structure preview	visualize_structure()

Limitations & Future Ideas

• Detection is heuristic — verify manually for real datasets.
• Graph Data requires edge-list format (source/target columns); adjacency matrices are classified as Data Matrix.
• Extension: detect genomic data (columns named A, T, C, G or single-character values from {A,T,C,G}).
• Extension: output a confidence score (0–100%) instead of High/Medium.