Data Science for Beginners - Python, Azure ML and Tableau

Set up your Python environment using Anaconda and Jupyter Notebook to run code, create visualizations with matplotlib and Seaborn, and explore data through cleaning, preprocessing, modeling, and analysis.

Learn to launch Jupyter Notebook from Anaconda, create and rename Python notebooks, navigate the local host, run code cells, and use key shortcuts to save and manage notebooks.

Explore variables in Python, set up notebooks and folders. Switch cells to markdown and learn variable naming rules and data types with practical examples.

Identify data types in datasets, including int, float, string, bool, and datetime, and see how release year, duration, rating, and movie title map to them using Python print and type.

Learn how lists work in Python, using square brackets to store multiple items and access them by zero-based and negative indexing with movie examples like The Matrix, Aladdin, and Interstellar.

Explore dictionaries in Python, using key–value pairs in curly braces to store movie details like title, release year, genre, and rating, and retrieve data by keys.

Explore tuples and sets in Python, highlighting immutable tuples and mutable sets, with examples of indexing, parentheses versus curly braces, and add or remove operations using movie data.

Explore arithmetic and comparison operators in Python for data filtering and conditioning. Learn equals, not equals, greater than, less than, and observe addition, multiplication, division, and exponentiation in code.

Master Python conditional statements with if, elif, and else through practical examples, including comparing numbers and classifying ages to create a new data column for data science.

Learn how data scientists use for loops to iterate over lists and strings, print each item, and apply conditions to control output.

Master combining for loops with if and else to filter movies by year in data science for beginners, using lists, dictionaries, and f-strings for clear output.

Define and call Python functions with def, print outputs, return values, and build examples from add numbers to filtering movies after a year using for loops and if statements.

Practice Python basics through five Q&A exercises, including dictionaries, conditional age classification, lists and for loops, number filtering, and a multiply function.

explore descriptive statistics for data summarization, focusing on mean, median, and mode, and learn how outliers and central tendency shape initial data insights.

Compute variance from squared deviations about the mean and derive standard deviation as its square root to gauge dispersion and risk in stock prices.

Explore sampling techniques in data science, including simple random, systematic, stratified, and cluster sampling, and learn how each method provides unbiased, representative conclusions from large populations.

Explore the two main variable types—quantitative and qualitative—and their subtypes, including discrete vs continuous and nominal vs ordinal, with concrete examples like height, salary, and performance ratings.

Explore frequency distributions to organize and summarize data, using ungrouped distributions for individual values and grouped distributions for large ranges like ages, with bar charts illustrating insights.

Learn to use pandas to read csv files into dataframes with pd.read_csv, load Titanic and salaries datasets, and view their top five and bottom five records.

Explore data frames with pandas, using describe to summarize numeric columns and reveal counts, means, stds, min, max, and quartiles; also inspect with head, tail, and sample.

Learn to perform algebraic operations in pandas by computing mean, min, max, sum, and std on columns, using dot notation or square brackets with a Titanic example.

Rename columns to tidy data frames using the rename method, such as changing work year to year and salary to annual salary, with inplace updates.

Preserve data integrity by creating checkpoint copies of dataframes, then use isnull and sum to identify missing values, and handle them with median imputation or selective column/row drops.

Explore data with python, aggregate using mean, sum, and count, group by fields, and use value counts to reveal trends in Titanic and salaries.

Use pandas groupby to group data by one or more columns, then apply max, mean, or value count, and sort to reveal top salaries.

Master filtering data in pandas with square brackets, ampersand operator, and pipe operator, and is in method for multiple values, using titanic and salaries data frames as examples.

Create an age category column on the Titanic data frame by applying a lambda-based classification to assign child, adult, and senior categories.

Convert the order date key to a date time field using to_datetime with a year-month-day format, then derive month and day as new features for time series analysis and modeling.

Plotting data with pandas: group by date or calendar year and month, compute total sales, and create line or bar plots to reveal trends.

practice pandas by loading Titanic dataset, renaming columns to gender and ticket_price, computing mean age by class, and displaying the names and ages of male passengers under 18 who survived.

Learn how to create and secure an OpenAI account, explore ChatGPT options (free 3.5 and paid GPT‑4), adjust settings, and prepare prompts to improve responses.

Assign a role to ChatGPT to set the context and tailor responses for a retail data analyst, surfacing insights on sales trends, customer behavior, and inventory management.

Learn to craft specific, actionable instructions for ChatGPT, break complex tasks into steps, and guide data tasks like calculating monthly sales figures and creating comparisons across categories.

Learn how providing rich context in prompts improves data analysis and visualizations, guiding insights with dataset details, stakeholders, goals, and contextual factors like demographics and regional trends.

Learn how few shot prompting improves model responses by using zero shot, one shot, and few shot examples to guide sentiment classification and output formats in tabular form.

Explore the limitations of ChatGPT, including lack of true understanding, biased outputs, hallucinations, and safety concerns, and learn to follow organizational guidelines when inputting data.

Use ChatGPT as a data analyst for an HR company to guide a pandas-based exploratory analysis of salaries data, including describe, info, and mean salaries by year.

Practice practical data analysis with ChatGPT by visualizing salary trends, including box plots, top paying titles, and location versus salary, with seaborn and matplotlib in Python.

Learn to create line plots with matplotlib and seaborn in Python, using pandas to read csv data, convert order dates to datetime, and visualize daily or monthly sales trends.

Explore data distribution with histograms to visualize salaries in USD, set bins, customize edge color, filter out outliers, and display plots with plt.show.

Learn to resize plots using plt.figure with figsize and dpi, setting a 10 by 6 size and 300 dpi for clearer visuals before adding plots.

Explore formatting options for histograms in matplotlib, including color, titles, labels, grid lines, and line styles, with ChatGPT-assisted code generation and seaborn themes.

Explore correlation and Pearson's coefficient, measuring linear relationships between two variables, with positive, negative, and no correlation, visual scatter plots, and notes on outliers and multicollinearity.

read the 2018 to 2019 world happiness dataset into a dataframe and build scatter plots of gdp per capita against healthy life expectancy to reveal the relationship across two years.

Create subplots to display multiple scatter plots in one figure, using grids like 1x2 or 2x2 with matplotlib, and prepare for seaborn plotting.

Create seaborn box plots from salary data to visualize the median, quartiles, and whiskers, highlighting data spread and outliers.

Violin plots reveal distribution shape with a density curve, showing how data spread varies across years like 2020, 2021, 2023, and 2024, and highlight the median and variability in salaries.

Visualize survival rates by passenger class in the Titanic dataset using Seaborn bar plots, with confidence intervals and options to explore gender and embarked categories.

Learn to create seaborn scatter plots, mapping GDP per capita to healthy life expectancy with hue by year, and explore three-metric visuals using happiness df and titanic df.

Display a correlation matrix for numeric columns in the Titanic data frame, selecting int64 and float64 types, and visualize it as a heat map with a customizable color map.

Learn to build a seaborn pair plot to visualize relationships among numerical columns in a happiness data frame, using hue by year and a two-color palette while dropping overall rank.

Create three Titanic data visualizations, a box plot of age by embarkation point, a histogram of age ranges, and a count plot of survivors by gender using seaborn and matplotlib.

Explore the machine learning lifecycle from identifying a business goal to maintaining a deployed model, learning how data processing, feature engineering, and model evaluation drive predictions.

Explore supervised and unsupervised learning, including labeled data and input features, to predict outcomes, discover patterns, and form clusters or topic groups.

Explore how linear regression predicts continuous outcomes like house prices from features such as square footage, and how logistic regression forecasts probabilities to classify binary outcomes like spam.

Explore unsupervised learning on unlabeled data by identifying patterns and forming clusters without predefined labels. Learn clustering techniques like k-means and hierarchical clustering to segment customers and power recommendations.

Practice a simple linear regression in Python with housing data, using square feet to predict price. Visualize the relationship with a scatter plot and a regression line.

Split data into training and test sets with an 80/20 split to train a linear regression model, evaluating generalization and avoiding overfitting in house price predictions from square feet.

Develop a multilinear regression model on the California housing prices data by loading the CSV, cleaning missing values, and visualizing the median house value as the target.

Learn to convert the ocean proximity category into numeric one-hot indicators with pandas Getdummies, creating binary columns, joining them to the housing data, and dropping the original feature.

Plot longitude and latitude in a seaborn scatter plot, color points by median house value with a blue–yellow palette, revealing a map of California and pricier homes near the ocean.

Explore how log transformations reduce skew in housing data features like total bedrooms and total rooms using numpy, producing more normal distributions for models like linear regression and random forests.

Apply feature engineering to create the average rooms per house feature by dividing total rooms by households, then assess its correlation with the median house value.

Understand multicollinearity and its impact on model interpretation by using a correlation matrix to spot highly correlated features and decide to remove or combine them for predicting median house price.

Display and interpret a correlation heatmap of housing data to reveal multicollinearity, create a bedroom ratio feature, drop redundant columns, and prepare data by splitting into training and testing sets.

Prepare x and y from housing data, perform a train-test split, and train a linear regression model. Fix infinite values from log transforms by adding one before evaluation.

Assess model performance with r squared, coefficient of determination, to see how well features explain variation in median house prices. Understand sums of squared errors and total sum of squares.

Explore mean squared error and root mean squared error to measure prediction accuracy using Python and scikit-learn, and compare models with different units or scales.

Explore how random forest regression uses an ensemble of decision trees trained on random samples and feature subsets to improve accuracy, reduce overfitting, and reveal key predictors.

Apply a random forest regressor to the housing data, compare to linear regression, and improve R-squared from 0.62 to about 0.82 with lower RMSE from ~71,000 to ~49,000.

Explore feature importance in a random forest regressor, identifying median income as the top predictor for median house price and visualizing results with a horizontal bar chart.

Hypothesis testing lets data scientists draw conclusions about populations from samples, testing assumptions to determine whether observed effects reflect signals or random chance, with null and alternative hypotheses introduced next.

Understand null and alternative hypotheses and how hypothesis testing uses sample data to decide whether to reject the null, illustrated with A/B tests and marketing scenarios.

Explore hypothesis tests with t tests and z tests to compare means across two groups, learn when to use small-sample t tests versus large-sample z tests, and interpret p value.

Explain the p value as a numerical measure of evidence against the null hypothesis. Use a 0.05 significance level to decide; illustrate with weight loss example in Jupyter notebook.

Compute the t statistic and p value for a two group t test in Python with SciPy, using a group scores data frame, and reject null when p below 0.05.

Explore signing up for Azure, claim a free 30-day trial with $200 credit, verify your email, set up multi-factor authentication, and start creating a resource group in the Azure portal.

Set up budgets and alerts in Azure Cost Management to monitor spending, receive email notifications at thresholds, and view cost analysis, invoices, and payment methods.

Create an Azure Machine Learning workspace and a compute cluster, then explore the designer and pipelines using DS 11 v2 compute in East US 2 for model training.

Create and import a loan data asset as a tabular dataset, load from local CSV, and explore with Azure ML profiling to identify features and loan approval label for modeling.

Build a numeric pipeline in azure machine learning designer, normalize data with min-max, split 80/20, train a two-class forest, and evaluate loan approval.

Explore how to evaluate a classification model using a confusion matrix, identifying true positives, true negatives, false positives, and false negatives in predictions to gauge performance.

Analyze model performance for binary classification using accuracy and area under the ROC curve (AUC) to distinguish between positive and negative classes, with ROC illustrating true and false positive trade-offs.

Explore how precision, recall, and F1 score evaluate model performance, illustrated with heart disease predictions, confusion matrix concepts, and handling imbalanced data.

Experience end-to-end model evaluation in Azure Machine Learning Studio by reviewing scored data, predicted labels, and probabilities, then interpret accuracy, precision, recall, F1, and AUC via the confusion matrix.