Data Science
Exam Study Guide
Pandas · Matplotlib · NumPy · Statistics · Clustering · GitHub
40 Questions · Quick Reference · Memory Tricks
■ EXAM FORMAT: 40 questions total — 20 from your 60 practice questions + 20 unseen. Master all 6 sections
below!
① NumPy — "Smart Lists for Math"
Think of NumPy arrays as Python lists that can do maths super fast.
Function What it does
[Link](5) Array of 5 zeros → [0,0,0,0,0]
[Link](5) Array of 5 ones → [1,1,1,1,1]
[Link](5, 7) Array of 5 sevens → [7,7,7,7,7]
[Link](0,10) Integers 0–9 (like range())
[Link](0,1,5) 5 evenly spaced values between 0 and 1
[Link](a) Copy array without linking to original
[Link](a) Sort array (ascending by default)
[Link](a,(2,3)) Change shape — same data, new layout
[Link]((a,b)) Stack vertically → adds ROWS
[Link]((a,b)) Stack horizontally → adds COLUMNS
[Link](a) Flip rows and columns
[Link](a) Running total: [1,2,3] → [1,3,6]
np.array_equal(a,b) True if both arrays are identical
■ Memory trick: vstack = Vertical = more Rows. hstack = Horizontal = more Columns.
② Pandas DataFrame — "Excel in Python"
A DataFrame is just a table with rows and columns. Every operation below works on that table.
Method / Function What it does
[Link]() Summary stats: mean, min, max, std, quartiles
[Link]() Delete ALL rows that contain any missing value
[Link](0) Fill missing values with 0 (or any value)
[Link](a, b) Replace every occurrence of a with b
df.reset_index() Reset row numbers back to 0, 1, 2 …
df.set_index("col") Use a column as the row label (index)
[Link]("col").mean() Group rows by col, then compute mean per group
df["new"] = df["a"]+df["b"] Add a new calculated column directly
[Link](new=...) Another way to add a new column (chainable)
[Link][row, col] Select by LABEL name
[Link][0, 1] Select by NUMBER position (row 0, col 1)
[Link](func) Apply a function to each row or column
df["col"].map(func) Apply a function to each CELL in one column
df["col"].value_counts() Count how many times each unique value appears
[Link](df1, df2) Join two DataFrames on a common key (SQL JOIN)
[Link]([df1,df2],axis=0) Stack DataFrames vertically (add rows)
[Link](df) Wide → Long: columns become rows
[Link]() Pivot innermost column level into row index
pd.pivot_table(df,...) Summarise like an Excel pivot table
■ loc vs iloc: loc = Label | iloc = Integer position
■ dropna REMOVES rows | fillna REPLACES missing values
■ merge = JOIN on key column | concat = just STACK tables
③ Matplotlib — "The Drawing Tool"
import [Link] as plt
fig, ax = [Link]() # Create figure + axes
[Link]() # Display the plot
Function / Parameter What it does
[Link]() Display / render the plot
[Link]("[Link]") Save plot to a file
[Link]() Create figure with one or more subplots
[Link]() Add background grid lines
[Link]() Draw a pie chart
[Link]() Draw a histogram
[Link](x=,y=) Scatter plot directly from a DataFrame
ax.set_title() Add a title to the axes
ax.set_xlim(a, b) Set x-axis visible range
ax.set_aspect("equal") Keep x and y scales proportional
ax.tick_params() Customise tick size, rotation, color
[Link]() Add annotation text with an arrow
[Link](x,y,"text") Add plain text at position (x,y)
ax.fill_between() Fill area between two curves
fig.set_size_inches() Change figure width and height
plt.tight_layout() Auto-fix spacing between subplots
subplots_adjust(wspace,hspace) Manually control subplot spacing
alpha=0.5 Transparency: 0=invisible, 1=fully solid
c=values Color scatter markers by data values
[Link]() Draw a box-and-whisker plot
④ Statistics — The 4 Moments + Distributions
The 4 Moments of a Distribution
Moment Name Simple meaning
1st Mean Average — the centre of the data
2nd Variance Spread — how far data sits from the mean
3rd Skewness Lean — left or right asymmetry
4th Kurtosis Tails — how heavy the extreme ends are
Key Distributions
Distribution Use case / Key idea
Normal Bell curve. Most natural real-world data.
Distribution Use case / Key idea
Uniform Every outcome equally likely (e.g. rolling a die).
Binomial Yes/No trials repeated n times. n = number of trials.
Poisson Count of events in a fixed time interval (calls/hour).
t-Distribution Like Normal but for SMALL sample sizes. Hypothesis testing.
Exponential Time between events.
■ Skewness: Positive → tail goes RIGHT (right-skewed) | Negative → tail goes LEFT (left-skewed)
■ Kurtosis of a Normal distribution = 3 (called Mesokurtic)
⑤ Fitting & Clustering Concepts
Fitting Clustering (K-Means)
Finding a mathematical line or curve that best describes Grouping data points that are similar — WITHOUT
your data points. • Linear fit = straight line through data • labels. • Choose K (number of clusters) • Algorithm
The goal is to minimise the error between the line and assigns each point to nearest centroid • Centroids move
actual points • Used in regression and prediction tasks until stable • Use Elbow Method to find best K
⑥ GitHub Operations — Version Control Basics
Command What it does
git init Start a brand new local repository
git clone Copy an existing repo to your machine
git add . Stage ALL changed files for commit
git commit -m "" Save a snapshot with a message
git push Upload your commits to GitHub
git pull Download latest changes from GitHub
git branch View or create branches
git merge Combine one branch into another
git status See which files have changed
git log View history of commits
■ Top Exam Tips — Read Before You Start
loc vs iloc loc = select by LABEL name | iloc = select by INTEGER position
dropna vs fillna dropna REMOVES rows with NaN | fillna REPLACES NaN with a value
merge vs concat merge = JOIN on a key column | concat = just STACK tables together
apply vs map apply works on DataFrame rows/columns | map works on Series elements
vstack vs hstack vstack → adds ROWS (vertical) | hstack → adds COLUMNS (horizontal)
Skewness sign Positive skew → right tail (right-skewed) | Negative → left tail
Kurtosis = 3 Normal distribution kurtosis = 3 = Mesokurtic
Poisson vs Binomial Poisson = events per time interval | Binomial = yes/no repeated n times
t-Distribution Use for hypothesis testing with SMALL sample sizes
alpha parameter Controls transparency in plots: 0 = invisible, 1 = solid
c parameter Colors scatter plot markers by data values
Unseen questions Read carefully — function names usually describe what they do!
Good luck tomorrow! You got this ■