8 End-to-End Data Analysis Project Free Online FlashCards

What should a data analysis project plan specify?

Define the question, unit of observation, target, features available at prediction time, and success criteria.

What should you inspect when first examining a table?

Check the table’s shape, sample rows, column types, unique values, duplicates, and missing values to catch problems early.

Why shouldn’t missing values automatically be replaced with zero?

A missing value may mean “not recorded,” “not applicable,” or something else; replacing it with zero can misrepresent its meaning.

What is the purpose of exploratory data analysis?

Exploratory data analysis uses summaries and visualizations to understand distributions, unusual observations, and possible relationships.

What do histograms, scatter plots, and bar charts commonly show?

Histograms show distributions, scatter plots show relationships between two numeric variables, and bar charts compare category counts.

What can’t a relationship in a plot establish by itself?

A plotted relationship or correlation indicates association, but does not establish that one variable causes another.

How do regression and classification differ?

Regression predicts a numeric target; classification predicts a category.

Why split off test data before fitting preprocessing?

Set aside the test set before fitting transformations or models, so test information cannot leak into training and inflate evaluation.

How does the example preprocess numeric and categorical features?

In the example, numeric features are median-imputed and standardized; the categorical feature is filled with its most frequent value and one-hot encoded.

What is a unit of observation?

The unit of observation is what one row represents, such as one customer.

Why might the median better summarize skewed data than the mean?

Extreme values can pull the mean, while the median is often more representative of skewed data.

What does mean absolute error (MAE) measure?

Mean absolute error is the average absolute prediction error, expressed in the target’s units; lower is better.