1. Objectives
- Set up a Python environment for AI work and check library versions.
- Create NumPy arrays, use indexing, slicing, broadcasting and vectorised maths (no loops).
- Use matrix multiplication (dot product) - the core operation of every neural network.
- Load a CSV file with pandas; select, filter, sort, group and add new columns.
- Draw and save line, scatter and bar plots with Matplotlib, with title, labels and legend.
- Compute a single artificial neuron's output z = X·w + b for many inputs at once.
2. Word Bank
Word | Simple meaning | Real-world example |
|---|---|---|
Library | Ready-made code you import and use | import numpy as np |
Array | A grid of numbers of the same type (1-D list, 2-D table) | Marks of 5 students in 3 quizzes |
Shape | Size of an array in each direction (rows, columns) | (5, 3) = 5 rows, 3 columns |
Vectorised | One operation on the whole array, without a loop | units * 35 for all houses at once |
Broadcasting | NumPy stretches a smaller array to match a bigger one | Adding one row of bonus marks to every student |
Dot product | Multiply matching numbers and add them up | Bill = units · rates |
DataFrame | A pandas table with named columns, like an Excel sheet | bills.csv loaded into df |
Boolean mask | An array of True/False used to pick rows | df[df['units'] > 300] |
Correlation | A number from -1 to +1 showing how two things move together | Hotter months → more units |
Neuron (artificial) | Weighted sum of inputs plus bias, then an activation | Good-yield predictor in Task 5 |
3. Background
Why these three libraries?
Almost every AI program in Python stands on three libraries. NumPy does fast maths on arrays: a neural network is mostly matrix multiplication, and NumPy does it in one line. pandas reads and cleans tables of data (CSV or Excel files), because AI learns from data. Matplotlib draws graphs, because we must see data and results before we trust them. scikit-learn and scikit-fuzzy, which we use in later labs, are built on NumPy.
Real-world link
A WAPDA/MEPCO bill is units × rate. If you have 1,000 houses, a loop calculates 1,000 bills one by one. NumPy calculates all 1,000 with one line: bills = units * rate. This is called vectorised (whole-array) maths, and it is about 50-100 times faster than a Python loop on large data.
Today's data
Two small files: bills.csv (monthly temperature and electricity units of a house in Burewala) and cotton.csv (10 cotton farms in Burewala, Vehari and Mailsi with area, rain, fertilizer bags and yield in maunds). Type them into Notepad / VS Code exactly as given in the box and save them in the same folder as your notebook.
Data files for this lab
bills.csv
month,avg_temp_c,units Jan,13,180 Feb,16,170 Mar,22,210 Apr,29,290 May,34,420 Jun,36,510 Jul,33,480 Aug,32,460 Sep,31,400 Oct,26,280 Nov,19,200 Dec,14,190 |
cotton.csv
farm,district,area_acres,rain_mm,fert_bags,yield_maunds F01,Burewala,5,310,10,160 F02,Burewala,8,280,14,232 F03,Vehari,4,350,6,96 F04,Vehari,6,300,12,186 F05,Mailsi,7,220,9,161 F06,Mailsi,3,240,4,63 F07,Burewala,10,330,20,330 F08,Vehari,5,260,7,120 F09,Mailsi,6,250,11,168 F10,Burewala,4,290,8,124 |
4. Time Plan (180 minutes)
Minutes | Activity | Part |
|---|---|---|
0-15 | Setup check, create data files | Part A1 |
15-60 | Guided warm-up: NumPy, dot product, pandas, plotting | Part A2-A5 |
60-75 | Predict the output | Part B |
75-95 | Bug hunt | Part C |
95-165 | Graded tasks 1-5 | Part D |
165-180 | Post-lab questions, lab book entry, submission | Part E |
Part A - Guided Warm-up (type, run, observe)
A1. Setup check
Install once: pip install numpy pandas matplotlib (Anaconda already has them). Then run this cell. Your version numbers may be different - any NumPy 1.26+ and pandas 2.0+ is fine.
import sys, numpy as np, pandas as pd, matplotlib print("Python ", sys.version.split()[0]) print("NumPy ", np.__version__) print("pandas ", pd.__version__) print("Matplotlib", matplotlib.__version__) |
Output Python 3.13.16 NumPy 2.5.3 pandas 3.0.5 Matplotlib 3.11.2 |
A2. Arrays and vectorised maths
Five houses on one street used these units last month. The rate is Rs 35 per unit. Compare the loop with the one-line NumPy version - both give the same bills.
import numpy as np units = np.array([120, 250, 310, 95, 480]) print("Units :", units) print("Shape :", units.shape, " Type:", units.dtype) # loop way bills_loop = [] for u in units: bills_loop.append(int(u * 35)) print("Loop :", bills_loop) # vectorised way (no loop) bills = units * 35 print("NumPy :", bills) print("Total Rs", bills.sum(), "| Average Rs", bills.mean(), "| Highest Rs", bills.max()) print("Houses above 300 units:", units[units > 300]) |
Output Units : [120 250 310 95 480] Shape : (5,) Type: int64 Loop : [4200, 8750, 10850, 3325, 16800] NumPy : [ 4200 8750 10850 3325 16800] Total Rs 43925 | Average Rs 8785.0 | Highest Rs 16800 Houses above 300 units: [310 480] |
A3. 2-D arrays and the dot product
A 2-D array is a table. The dot product multiplies matching numbers and adds them. Here each row is a student, each column a quiz; the weights say how much each quiz counts. One @ (matrix multiply) gives every student's weighted score.
import numpy as np marks = np.array([[8, 7, 9], [5, 6, 4], [9, 9, 10], [6, 8, 7]]) weights = np.array([0.2, 0.3, 0.5]) # quiz 3 counts most print("Shape of marks :", marks.shape) print("Student 2 marks :", marks[1]) print("Quiz 3 column :", marks[:, 2]) weighted = marks @ weights # same as np.dot(marks, weights) print("Weighted scores :", weighted.round(2)) print("By hand, student 1: 8*0.2 + 7*0.3 + 9*0.5 =", round(8*0.2 + 7*0.3 + 9*0.5, 2)) |
Output Shape of marks : (4, 3) Student 2 marks : [5 6 4] Quiz 3 column : [ 9 4 10 7] Weighted scores : [8.2 4.8 9.5 7.1] By hand, student 1: 8*0.2 + 7*0.3 + 9*0.5 = 8.2 |
A4. pandas: load, look, filter, sort
pandas reads the CSV into a DataFrame (a table). Each column has a name, so we write df['units'] instead of remembering column numbers.
import pandas as pd df = pd.read_csv("bills.csv") print(df.head(3)) print("Rows, columns:", df.shape) print("Average units:", round(df["units"].mean(), 1)) hot = df[df["avg_temp_c"] >= 30] print("Hot months:", list(hot["month"])) print(df.sort_values("units", ascending=False).head(3)) |
Output month avg_temp_c units 0 Jan 13 180 1 Feb 16 170 2 Mar 22 210 Rows, columns: (12, 3) Average units: 315.8 Hot months: ['May', 'Jun', 'Jul', 'Aug', 'Sep'] month avg_temp_c units 5 Jun 36 510 6 Jul 33 480 7 Aug 32 460 |
A5. Matplotlib: line plot saved to a file
Always give a plot a title, axis labels with units, and save it with savefig BEFORE plt.show().
import pandas as pd import matplotlib.pyplot as plt df = pd.read_csv("bills.csv") plt.figure(figsize=(7, 3.5)) plt.plot(df["month"], df["units"], marker="o", label="Units used") plt.title("Monthly electricity use - house in Burewala") plt.xlabel("Month"); plt.ylabel("Units (kWh)") plt.grid(alpha=0.3); plt.legend() plt.tight_layout() plt.savefig("bills_plot.png", dpi=100) print("Saved bills_plot.png") |
Output Saved bills_plot.png |
Part B - Predict the Output
Write your prediction BEFORE running the code. Then run it and correct yourself in a different colour.
B1.
import numpy as np a = np.arange(1, 7) print(a.reshape(2, 3)) print(a.reshape(2, 3).shape) |
My prediction |
B2.
import numpy as np a = np.array([10, 20, 30, 40, 50]) print(a[1:4], a[-1], a[::2]) |
My prediction |
B3.
import numpy as np m = np.array([[1, 2], [3, 4]]) print(m * 10) print(m + np.array([100, 200])) |
My prediction |
B4.
import numpy as np t = np.array([28, 36, 41, 33, 39]) print(t[t > 35]) print((t > 35).sum()) |
My prediction |
B5.
import numpy as np m = np.array([[2, 4], [6, 8]]) print(m.sum(), m.sum(axis=0), m.sum(axis=1), m.mean()) |
My prediction |
Part C - Bug Hunt
Each program has ONE bug. Some crash with an error message; others run but give a wrong answer (silent logic error - the dangerous kind). Find the bug, explain it and write the fix.
C1.
import numpy as np a = np.array([1, 2, 3]) b = np.array([4, 5]) print(a + b) |
Bug, why it happens, and the fixed line(s) |
C2.
import pandas as pd df = pd.read_csv("bills.csv") print(df["Units"].max()) |
Bug, why it happens, and the fixed line(s) |
C3.
import pandas as pd df = pd.read_csv("bills.csv") print(df[df["units"] > 300 and df["avg_temp_c"] > 33]) |
Bug, why it happens, and the fixed line(s) |
C4.
import numpy as np temps = np.array([30, 32, 35]) backup = temps # keep the original safe temps[0] = 99 print("backup:", backup) # we expect [30 32 35] |
It runs and prints backup: [99 32 35] |
Bug, why it happens, and the fixed line(s) |
C5.
import numpy as np marks = np.array([[8, 7, 9], [5, 6, 4], [9, 9, 10], [6, 8, 7]]) quiz_avg = marks.mean(axis=1) # average of each quiz print("Quiz averages:", quiz_avg) |
It runs and prints Quiz averages: [8. 5. 9.33333333 7. ] |
Bug, why it happens, and the fixed line(s) |
Part D - Graded Tasks (20 marks)
Task | Level | Title | Marks |
|---|---|---|---|
T1 | Easy | Weekly temperature report | 3 |
T2 | Medium | Quiz marks analysis | 4 |
T3 | Medium | Electricity bills with pandas | 4 |
T4 | Medium | Plotting the bills | 4 |
T5 | Challenge | One artificial neuron (preview of Practical 2) | 5 |
Total | 20 |
Task 1 (Easy, 3 marks): Weekly temperature report
The maximum temperatures (°C) in Burewala for one week in June were 38, 41, 43, 40, 36, 44, 42. Store them in a NumPy array and print: the highest, the lowest, the average (1 decimal), how many days were above 40 °C, and which day numbers (1-7) were above 40 °C. Use NumPy functions, no loops.
Sample run (your output must match):
Output Highest : 44 Lowest : 36 Average : 40.6 Days above 40: 4 Day numbers : [2 3 6 7] |
Your code (or attach the file) |
Task 2 (Medium, 4 marks): Quiz marks analysis
Marks (out of 10) of 5 students in 3 quizzes are: Ali [7, 8, 6], Sana [9, 9, 10], Bilal [4, 6, 5], Hira [8, 7, 9], Usman [6, 5, 7]. Store them in a 5×3 array. Print (a) each student's total, (b) the average of each quiz (2 decimals), (c) the name of the top student (use argmax), and (d) the marks scaled to 0-1 with min-max scaling: (x - min) / (max - min) of the whole array, rounded to 2 decimals.
Sample run (your output must match):
Output Totals : [21 28 15 24 18] Quiz average: [6.8 7. 7.4] Top student : Sana Scaled 0-1: [[0.5 0.67 0.33] [0.83 0.83 1. ] [0. 0.33 0.17] [0.67 0.5 0.83] [0.33 0.17 0.5 ]] |
Your code (or attach the file) |
Task 3 (Medium, 4 marks): Electricity bills with pandas
Load bills.csv. (a) Add a column 'bill' using a simple slab tariff: Rs 22 per unit if units ≤ 200, otherwise Rs 35 per unit for ALL units (use np.where). (b) Print the months with units above 400 and their bills. (c) Add a 'season' column: 'Summer' for Apr-Sep, else 'Winter', and print the average units of each season with groupby. (d) Print the total yearly bill.
Sample run (your output must match):
Output month units bill 4 May 420 14700 5 Jun 510 17850 6 Jul 480 16800 7 Aug 460 16100 season Summer 426.7 Winter 205.0 Name: units, dtype: float64 Yearly bill Rs 123030 |
Your code (or attach the file) |
Task 4 (Medium, 4 marks): Plotting the bills
Using bills.csv, make ONE figure with two side-by-side plots (plt.subplots(1, 2)): left - a bar chart of units per month; right - a scatter plot of avg_temp_c (x) against units (y). Give each plot a title and axis labels, save the figure as task4.png, and print the correlation between temperature and units (df['avg_temp_c'].corr(df['units']), 3 decimals). In one line, explain what the correlation means.
Sample run (your output must match):
Output Saved task4.png Correlation: 0.937 Strong positive: hotter months use more electricity (fans and ACs). |
Your code (or attach the file) |
Task 5 (Challenge, 5 marks): One artificial neuron (preview of Practical 2)
Load cotton.csv. Make two input features for each farm: x1 = fert_bags / area_acres (bags per acre) and x2 = rain_mm / 100. Stack them into a 10×2 array X. Label a farm 'good' (1) if yield per acre ≥ 28 maunds, else 0. A neuron with weights w = [1.5, 0.5] and bias b = -4.5 computes z = X @ w + b and predicts 1 if z ≥ 0 (step activation). (a) Print z (3 decimals), predictions and true labels. (b) Print the accuracy. (c) Print the farm names it got wrong. (d) Try w = [2.0, 0.5], b = -5.2 and print the new accuracy. No loops are allowed.
Sample run (your output must match):
Output z : [ 0.05 -0.475 -0.5 0. -1.471 -1.3 0.15 -1.1 -0.5 -0.05 ] pred : [1 0 0 1 0 0 1 0 0 0] true : [1 1 0 1 0 0 1 0 1 1] Accuracy: 0.7 Wrong farms: ['F02', 'F09', 'F10'] New accuracy: 0.8 |
Your code (or attach the file) |
Part E - Post-Lab Questions
E1. Why is vectorised NumPy code preferred over Python loops in AI programs?
Your answer |
E2. What is the difference between axis=0 and axis=1 in a 2-D array? Give an example from Task 2.
Your answer |
E3. Write the formula a single artificial neuron uses, and name each part.
Your answer |
E4. Why should you save a figure before calling plt.show()?
Your answer |
E5. Task 4 found a correlation of 0.937 between temperature and units. Does heat CAUSE the units? Explain.
Your answer |
Marking Rubric (applied to Part D)
Criterion | Full | Partial | None |
|---|---|---|---|
Code runs without errors | Runs fully | Small fixable error | Does not run |
Correct output | All parts match | Most parts | Few or none |
Uses NumPy/pandas properly (no loops where asked) | Vectorised throughout | Some loops | Loops everywhere |
Plots: title, labels, saved file | All present | Some missing | No plot |
Code style: names and comments | Clear | Partly clear | Unclear |
Submission
- Make a folder RollNo_P01 (e.g. 2023-ag-1234_P01) with task1.py ... task5.py (or one notebook P01.ipynb), bills.csv, cotton.csv and task4.png.
- Zip the folder and upload it to the LMS / class Google Classroom by the end of the lab day.
- Copy Task 2 and Task 5 code and output into your Lab Book (Practical 1 pages) for the instructor's signature next week.
- Copied code from another student gets zero for both students. Using AI tools to explain errors is allowed; say so in a comment.