Learn Python from the beginning, with small examples drawn from data, machine learning, and everyday programming.

An underwater microphone recorded five simplified frequency readings overnight. Run the detector to see whether any reading falls inside the review range.

Real acoustic analysis uses frequency, duration, repetition, background noise, and labeled recordings. This small program gives us a useful first look at the shape of the problem.

A simplified detector

At 4:36 AM, an underwater microphone sixty kilometers offshore uploads five simplified frequency readings. One of them is much lower than the surrounding noise.

You do not need to understand every line yet. Run the detector and see whether it notices the unusual reading.

The reading at 42 falls below the review limit. Look at the final line in the output after you press Run.

What to watch

The overnight recording

The reading at 42 triggered the detector. Replace 42 with 110, then run the program again.

The code stays the same, but the conclusion changes with the recording. That is the first important idea: a program can respond to its data instead of memorizing one answer.

Change the recording

The detector currently reviews frequencies below 60. Change whale_range to 30. The reading at 42 should no longer be flagged.

Run the code when you are ready. You will see whether both the setting and the result are correct.

Tune the listening range

The final part gives the program two possible paths. If matches contains a reading, Python reports a possible call. Otherwise, it clears the recording.

You are not expected to write this decision from memory yet. Later lessons will unpack the list, comparison, and condition one piece at a time.

You will return to this

Follow the decision

A new recording contains a low reading at 38. Complete the missing report so the detector can flag it for review.

A second recording arrives

Your first program received observations, searched for a matching value, applied a threshold, made a decision, and reported the result.

A real whale call detector considers far more than one frequency threshold. The important part today is seeing how data can move through code and produce a decision.

A first look, not a final model

The next recording will contain different values. To work with changing information, the program needs a way to name and store it. That is where variables come in.

The shape of the problem

Run and modify a tiny detector that flags a possible whale call in underwater frequency readings.

Your first Python program

A variable gives a name to a value. Use clear names so another person can understand what the value represents.

Imagine seeing the number 0.92 by itself. You would have to guess what it means. A name such as confidence_score gives the number context. Good variable names are especially useful in AI code because the same program may contain inputs, predictions, labels, scores, and settings.

Use lowercase words separated by underscores, such as favorite_color or lesson_count.

Naming variables

Describe a training run

Store information in variables and inspect the type of each value.

Variables and values

A string is a sequence of characters. Python gives you simple tools for joining text, changing its case, and selecting part of it.

An f string places values inside text with braces. Indexes start at zero, so the first character is at index 0. A slice selects a range of characters.

Dataset labels, file names, prompts, model names, and generated answers are all strings. You will often clean them with strip, compare them in lowercase, or combine them into a readable message. None of that is advanced, but messy text causes plenty of real bugs.

String methods return a new string. Save that result in a variable if you want to use it later.

Strings cannot be changed in place

Format a prediction

Combine, format, slice, and transform text.

Working with strings

Python supports addition, subtraction, multiplication, division, powers, and remainders. Parentheses make the order of a calculation clear.

The word metric can sound more complicated than it is. Accuracy, for example, is the number of correct predictions divided by the total number of predictions. Python does not need to know that the result is an AI metric. It simply follows the arithmetic you write.

Use int, float, and str to convert compatible values from one type to another.

Converting values

Calculate model accuracy

Calculate with integers and floats, then learn how operator order works.

Numbers and math

A comparison produces either True or False. Your program can use that result to decide what happens next.

Use two equals signs to compare values. A single equals sign assigns a value. You can combine comparisons with and, or, and not.

Many models return a score rather than a final yes or no answer. Your Python code chooses a threshold and turns that score into a useful decision. The model may produce 0.81, while an if statement turns it into spam, fraud, positive, or whatever label your application needs.

Python normally uses four spaces for each indentation level.

Indentation matters

Apply a confidence threshold

Compare values, combine conditions, and choose which code should run.

Booleans and decisions

A list stores an ordered collection that can change. A tuple stores an ordered collection that is meant to stay fixed.

Negative indexes count from the end. A slice can select several items without changing the original list.

A dataset is ultimately a collection of examples. A model often returns a collection of predictions. Lists are the first Python structure that lets you work with those groups naturally. You can add an item, inspect one position, or pass the whole list to a function.

Start with a list. Use a tuple when the fixed structure of the value is important.

Choose the simple option

Summarize prediction scores

Store ordered groups of values and select the items you need.

Lists and tuples

A dictionary connects each key to a value. A set stores unique values and makes membership checks easy.

A row of data often looks like a dictionary

A dictionary is a natural way to represent one example because every value has a name. A prediction record might contain the input identifier, predicted label, and confidence. Sets are handy when you only care which labels appear, not how many times each one appears.

Build a prediction record

Look up values by name and keep collections of unique items.

Dictionaries and sets

A loop repeats a block of code. Use a for loop to visit items in a collection. Use a while loop when repetition depends on a condition.

The enumerate function gives you both the position and the value. The range function is useful when you need a sequence of numbers.

Loops let you inspect a dataset one item at a time

Libraries can process large arrays for you later. For now, a normal loop is the clearest way to see what repeated work means. You take one score, do something with it, and move to the next score until the collection is finished.

A while loop must eventually make its condition false.

Avoid endless loops

Count confident predictions

Turn scores into labels

Repeat work with for loops and while loops.

Loops

A function gives a name to a reusable piece of behavior. Parameters receive input, and return sends a result back to the caller.

A line such as classify_score(0.82) is easy to read because the function name explains the intention. This becomes important as an AI project grows. Loading data, cleaning text, making a prediction, and calculating a metric should not become one enormous script.

If you copy the same logic several times, it may belong in a function.

Functions reduce repetition

Write a score classifier

Calculate mean absolute error

Group reusable instructions and return a result.

Writing functions

Errors are part of programming. Read the final line of the error message first. It usually names the error and explains what went wrong.

Data is rarely as tidy as the example in a tutorial. A blank field, unexpected word, or missing value can break an otherwise reasonable program. Handling the specific problem you expect lets the program continue without hiding unrelated mistakes.

Avoid catching every possible error. A specific exception makes unexpected problems visible.

Catch specific errors

Clean a mixed batch

Read error messages, protect risky code, and find simple mistakes.

Errors and debugging

A module is a Python file that contains useful values and functions. The standard library includes modules for math, dates, random values, files, and much more.

Use import module when you want the module name to show where a function came from. Use from module import name for a small number of frequently used names.

NumPy, pandas, and scikit learn are external packages, but you import them with the same syntax. Learning imports with the standard library first means there is one less new idea to absorb when you eventually load a real dataset or model.

External packages are installed separately. You will meet them after the Python basics feel comfortable.

The standard library comes with Python

Sample a data point

Use code from Python modules instead of writing everything yourself.

Modules and imports

Files let a program keep information after it stops running. Use with open so Python closes the file correctly when the block finishes.

A real data workflow often reads input from one file and saves results to another. More specialized formats such as CSV and JSON come later, but the basic rhythm is already here: open, read or write, and let the with block close the file safely.

You can now read and write small Python programs. The next step is practice, not memorizing every corner of the language.

Your Python Basics journey