Page | 1

Author: Muhammad Omar Akhlaq

 1. PANDAS
Data Reading and Writing:

 read_csv(): Reads data from a CSV file into a DataFrame.

 read_excel(): Reads data from an Excel file into a DataFrame.
 to_csv(): Writes data from a DataFrame to a CSV file.
 to_excel(): Writes data from a DataFrame to an Excel file.

DataFrame Basics:

 head(): Displays the first few rows of the DataFrame.

 tail(): Displays the last few rows of the DataFrame.
 info(): Provides concise summary information about the DataFrame.
 describe(): Generates descriptive statistics of the DataFrame.
 shape: Returns the dimensions (rows, columns) of the DataFrame.
 columns: Returns the column labels of the DataFrame.
 index: Returns the row labels of the DataFrame.

Data Selection and Manipulation:

 loc[]: Accesses a group of rows and columns by labels.

 iloc[]: Accesses a group of rows and columns by integer position.
 drop(): Drops specified rows or columns from the DataFrame.
 fillna(): Fills NaN (missing) values in the DataFrame with specified values.
 groupby(): Groups data based on specified criteria.
 merge(): Merges DataFrames using a database-style join operation.

Data Aggregation and Calculation:

 sum(), mean(), median(): Aggregate functions for calculations.

 max(), min(): Finds maximum and minimum values.
 apply(): Applies a function along an axis of the DataFrame.
 pivot_table(): Creates a spreadsheet-style pivot table as a DataFrame.

Data Cleaning and Handling:

 drop_duplicates(): Drops duplicate rows from the DataFrame.

 rename(): Renames columns or index labels.
 astype(): Converts the data type of a column to another data type.
 set_index(), reset_index(): Sets or resets the DataFrame index.
 Time Series Analysis (for date-time data):
 to_datetime(): Converts a column to datetime format.
 resample(): Performs data aggregation at different time frequencies.

Page | 2
Author: Muhammad Omar Akhlaq
  date_range(): Generates date-time indices.

2. NUMPY
Array Creation:

 np.array(): Creates an array from a Python list or tuple.

 np.zeros(): Generates an array of zeros with a specified shape.
 np.ones(): Generates an array of ones with a specified shape.
 np.arange(): Creates an array with evenly spaced values within a given interval.
 np.linspace(): Generates an array with a specified number of elements within a range.

Array Manipulation:

 np.reshape(): Changes the shape of an array without changing its data.

 np.flatten(), np.ravel(): Flattens a multidimensional array into a 1D array.
 np.concatenate(): Joins arrays along a specified axis.
 np.split(): Splits an array into multiple sub-arrays.
 np.transpose(), ndarray.T: Transposes array dimensions.
 np.vstack(), np.hstack(): Stacks arrays vertically or horizontally.

Array Operations:

 np.sum(), np.mean(), np.median(): Aggregate functions for calculations.

 np.max(), np.min(): Finds maximum and minimum values.
 np.dot(): Performs matrix multiplication.
 np.sort(): Sorts elements in an array.
 np.unique(): Finds unique elements in an array.

Array Indexing and Slicing:

 Slicing: Allows accessing sub-arrays within arrays using indices.

 Fancy Indexing: Uses arrays of indices to access specific elements.
 Mathematical Functions:
 np.sin(), np.cos(), np.tan(): Trigonometric functions.
 np.exp(), np.log(), np.sqrt(): Exponential, logarithmic, and square root functions.
 np.absolute(), np.power(): Absolute value and exponentiation functions.

Random Sampling:

 np.random.rand(): Generates random numbers from a uniform distribution.

 np.random.randn(): Generates random numbers from a normal distribution.
 np.random.randint(): Generates random integers within a specified range.
 np.random.choice(): Picks random elements from an array.

Linear Algebra:
Page | 3
Author: Muhammad Omar Akhlaq
  np.linalg.inv(): Computes the inverse of a matrix.
 np.linalg.det(): Computes the determinant of a matrix.
 np.linalg.solve(): Solves a system of linear equations.

3. SciKit-Learn
Data Preprocessing:
 StandardScaler: Standardizes features by removing the mean and scaling to unit variance.
 MinMaxScaler: Scales features to a given range (typically 0 to 1).
 OneHotEncoder, LabelEncoder: Converts categorical variables into numerical representations.
 train_test_split: Splits datasets into training and testing subsets for model evaluation.

Supervised Learning:
Regression:
 LinearRegression, Ridge, Lasso, ElasticNet: Linear regression and regularized versions.
 Classification:
 LogisticRegression, SVC, RandomForestClassifier, KNeighborsClassifier: Classifiers for different
algorithms.
 DecisionTreeClassifier, GradientBoostingClassifier: Decision tree-based models.

Unsupervised Learning:
Clustering:
 KMeans, DBSCAN, AgglomerativeClustering: Algorithms for clustering data.
 Dimensionality Reduction:
 PCA, TruncatedSVD, FactorAnalysis: Methods for reducing dimensions while preserving important
information.

Model Evaluation and Metrics:

 accuracy_score, precision_score, recall_score: Evaluation metrics for classification models.
 mean_squared_error, mean_absolute_error, r2_score: Evaluation metrics for regression models.
 cross_val_score, GridSearchCV: Tools for cross-validation and hyperparameter tuning.
 confusion_matrix, classification_report: Metrics to assess classification performance.

Pipelines and Feature Selection:

 Pipeline: Chains together multiple steps into a single workflow.
 FeatureUnion: Combines different transformers applied in parallel.
 SelectKBest, SelectFromModel: Feature selection methods based on statistics or model importance.

Ensemble Methods:
 VotingClassifier, VotingRegressor: Combines multiple models' predictions for improved performance.
 BaggingClassifier, RandomForestClassifier: Bootstrap aggregating methods for classification.
 GradientBoostingClassifier, AdaBoostClassifier: Boosting methods for classification.

Neural Network Support:

 MLPClassifier, MLPRegressor: Multi-layer Perceptron models for classification and regression.

Page | 4
Author: Muhammad Omar Akhlaq
 Text Analysis:
 CountVectorizer, TfidfVectorizer: Convert text data into numerical feature vectors.
 TfidfTransformer: Applies Term Frequency-Inverse Document Frequency transformation.

Model Serialization:
 pickle: Built-in Python module for serializing and deserializing scikit-learn models.

4. TENSOR FLOW
Core Components:
 Tensors: Fundamental data structures in TensorFlow, similar to multi-dimensional arrays.
 Operations: Functions that manipulate tensors, perform computations, and define the computational
graph.
 Graph: The computational graph defines the operations and dependencies between tensors.

Model Building:
 Keras API: High-level API within TensorFlow for building and training neural networks easily.
 tf.keras.layers: Module containing various layers like Dense, Conv2D, LSTM, etc., for building neural
network architectures.
 tf.keras.Sequential: Allows the sequential stacking of layers to create a model.

Training and Optimization:

 tf.keras.Model.compile(): Compiles the model, defining loss functions, optimizers, and metrics.
 tf.keras.Model.fit(): Trains the model on training data.
 tf.keras.Model.evaluate(): Evaluates the model's performance on a test dataset.
 tf.optimizers: Module containing optimization algorithms like SGD, Adam, RMSprop, etc.

Customization and Extensions:

 tf.GradientTape: Records operations for automatic differentiation and custom gradient computation.
 tf.function: Converts Python functions into graph-based TensorFlow functions for performance
optimization.
 Custom Layers, Metrics, Losses, and Callbacks: Allows the creation of custom components for specific
needs.

Deployment and Serialization:

 tf.saved_model: Tools for saving and loading models in the SavedModel format for deployment.
 tf.keras.Model.save(), tf.keras.models.load_model(): Saving and loading models using the Keras API.

Data Handling:
 tf.data.Dataset: A powerful API for creating input pipelines to handle large datasets efficiently.
 tf.data.experimental: Module containing experimental features for data pipeline handling.

GPU and Distributed Computing:

 tf.device(): Context manager to explicitly specify the device (CPU or GPU) for execution.
 tf.distribute: Module providing tools for distributed training across multiple devices or machines.

Page | 5
Author: Muhammad Omar Akhlaq
 Miscellaneous:
 tf.math: Module containing mathematical operations on tensors.
 tf.image: Module for image processing operations in TensorFlow.
 tf.strings: Module for string manipulation operations.
5. KERAS
Model Building:
 Sequential: A linear stack of layers for building sequential models.
 Functional API (tf.keras.Model): Allows creating complex models with shared layers and multiple inputs
or outputs.
 Dense, Conv2D, LSTM, Dropout, etc.: Various layer types for constructing neural network architectures.

Compilation and Configuration:

 compile(): Configures the model for training by specifying the optimizer, loss function, and metrics.
 Optimizers (SGD, Adam, RMSprop, etc.): Algorithms for optimizing model weights during training.
 Loss Functions (mean_squared_error, categorical_crossentropy, etc.): Measures the model's
performance.

Training and Evaluation:

 fit(): Trains the model on training data.
 evaluate(): Evaluates the model's performance on a test dataset.
 predict(): Generates predictions for new data.
 Callbacks (EarlyStopping, ModelCheckpoint, etc.): Tools for customizing training behavior.

Regularization and Optimization:

 Dropout, BatchNormalization: Techniques for regularization and improving training convergence.
 Regularizers: Methods for applying penalties on layer parameters during optimization.

Customization and Extension:

 Layer and Model classes: Allows building custom layers and models by subclassing.
 Custom Callbacks, Custom Losses, Custom Metrics: Enables customizing and extending Keras
functionality.

Serialization and Deployment:

 save() and load_model(): Functions for saving and loading models.
 model_to_json() and model_from_json(): Serialization of model architecture to/from JSON format.

Preprocessing and Utilities:

 preprocessing: Module for data preprocessing techniques like normalization, text tokenization, etc.
 utils: Module containing utility functions for working with Keras models and layers.

GPU and Distributed Computing:

 Keras models can leverage TensorFlow's capabilities for GPU and distributed computing seamlessly.

Page | 6
Author: Muhammad Omar Akhlaq
 6. PyTorch
Tensor Operations:
 torch.Tensor: The core data structure, supports various tensor operations similar to NumPy arrays.
 Math operations: Element-wise operations, matrix multiplications, and other mathematical functions.
 Indexing and Slicing: Accessing and manipulating tensor elements.
 Autograd and Dynamic Computation Graph:
 torch.autograd: Automatic differentiation engine for computing gradients.
 torch.autograd.Function: Base class for defining custom autograd operations.
 backward(): Computes gradients of tensors with respect to a given computational graph.

Neural Network Building Blocks:

 torch.nn: Module for building neural network architectures.
 torch.nn.Module: Base class for creating custom neural network modules.
 Layers: Various layers like Linear, Conv2d, LSTM, Dropout, etc.
 Activation functions: ReLU, Sigmoid, Tanh, etc.
 Initialization methods: Different weight initialization techniques.

Optimizers and Loss Functions:

 Optimizers (SGD, Adam, RMSprop, etc.): Algorithms for optimizing model weights.
 torch.optim: Module containing optimization algorithms.
 Loss functions (MSE Loss, CrossEntropyLoss, etc.): Measures the model's performance.

Training and Evaluation:

 torch.nn.functional: Functions for implementing neural network operations.
 torch.utils.data: Tools for handling datasets and data loaders for efficient batching and loading.
 torch.utils.data.Dataset: Base class for creating custom datasets.
 torch.utils.data.DataLoader: Loads data into batches for training and evaluation.

GPU and Distributed Computing:

 torch.device(): Context manager to specify the device (CPU or GPU) for tensor computations.
 torch.distributed: Module for distributed computing across multiple devices or machines.
 Serialization and Deployment:
 torch.save() and torch.load(): Functions for saving and loading models.
 Model serialization and deployment mechanisms for inference in production environments.

Miscellaneous:
 torch.cuda: Module for GPU-related functionalities and operations.
 torchvision: Module containing datasets, model architectures, and image transformation utilities for
computer vision tasks.
 torchtext: Module for text-related utilities and datasets.
Page | 7
Author: Muhammad Omar Akhlaq
 7. MatPlotLib
Basic Plotting:
 plt.plot(): Creates line plots.
 plt.scatter(): Generates scatter plots.
 plt.bar(), plt.barh(): Creates vertical and horizontal bar plots.
 plt.hist(): Generates histograms.
 plt.boxplot(): Creates boxplots to visualize data distributions.

Customization and Styling:

 plt.xlabel(), plt.ylabel(): Sets labels for the x and y axes.
 plt.title(): Sets the title of the plot.
 plt.legend(): Adds legends to the plot.
 plt.grid(): Displays grid lines on the plot.
 plt.xlim(), plt.ylim(): Sets the limits of the x and y axes.

Subplots and Layouts:

 plt.subplots(): Creates multiple subplots within a single figure.
 plt.subplot(): Adds individual subplots to a figure with custom configurations.
 plt.tight_layout(): Automatically adjusts subplot parameters for better layout.

Advanced Plot Types:

 plt.contour(), plt.contourf(): Generates contour plots.
 plt.pcolor(), plt.pcolormesh(): Creates pseudocolor plots.
 plt.quiver(): Displays vector fields.
 plt.imshow(): Displays images.

Annotations and Text:

 plt.text(): Adds text at specified coordinates on the plot.
 plt.annotate(): Annotates a specific point on the plot with optional arrow indicators.

Save and Show:

 plt.show(): Displays the plot.
 plt.savefig(): Saves the plot as an image file (PNG, JPG, PDF, etc.).

Specialized Plots:
 plt.pie(): Generates pie charts.
 plt.stem(): Creates stem plots.
 plt.violinplot(): Displays violin plots.
 3D Plotting (with mpl_toolkits.mplot3d):

Page | 8
Author: Muhammad Omar Akhlaq
  Axes3D: Provides 3D axes for plotting.
 plot_surface(): Plots 3D surfaces.
 scatter(): Creates 3D scatter plots.

Interactive and Animation:

 Interactive mode: Enables interactive plotting in supported environments.
 Animation: Module for creating animated plots.
8. Seaborn
Data Visualization:
 sns.lineplot(): Generates line plots with optional estimation and confidence intervals.
 sns.scatterplot(): Creates scatter plots with optional hue and size mapping.
 sns.barplot(), sns.countplot(): Produces bar plots and count plots.
 sns.boxplot(), sns.violinplot(): Displays boxplots and violin plots to show data distributions.

Categorical Data:
 sns.catplot(): Creates categorical plots (scatter, strip, box, violin, etc.) based on data types.
 sns.swarmplot(): Visualizes categorical data along with the distribution of observations.

Distribution Visualization:
 sns.histplot(), sns.kdeplot(): Displays histograms and kernel density estimation plots.
 sns.rugplot(): Shows individual data points as dashes on a plot axis.

Relationship Plots:
 sns.pairplot(): Creates a matrix of scatterplots for examining pairwise relationships in a dataset.
 sns.heatmap(): Generates a heatmap to visualize matrix-like data.

Regression and Model Visualization:

 sns.regplot(), sns.lmplot(): Displays linear regression models.
 sns.residplot(): Plots the residuals of a linear regression model.
 sns.jointplot(): Visualizes the relationship between two variables and their individual distributions.

Color Palettes and Styles:

 sns.set_palette(): Sets the color palette for the plot.
 sns.set_style(): Sets the visual aesthetic styles for the plot.
 sns.color_palette(): Creates color palettes for use in plots.

Axis and Figure-Level Functions:

 FacetGrid and PairGrid: Allow the creation of customized subplots.
 sns.relplot(), sns.lineplot(): Higher-level interfaces to create relational plots.

Themes and Contexts:

 sns.set_theme(): Sets the overall visual theme for the plots.
 sns.plotting_context(): Controls the scaling of plot elements.

Statistical Estimation:
Page | 9
Author: Muhammad Omar Akhlaq
  sns.pointplot(): Visualizes point estimates and confidence intervals.
 sns.barplot(): Displays the central tendency of a numeric variable.

9. Ploty
Basic Plotting:
 go.Scatter(): Generates scatter plots.
 go.Bar(): Creates bar charts.
 go.Histogram(): Displays histograms.
 go.Box(): Generates box plots.
 go.Surface(): Generates 3D surface plots.

Additional Plot Types:

 go.Pie(): Generates pie charts.
 go.Candlestick(): Displays financial candlestick charts.
 go.Heatmap(): Creates heatmaps.
 go.Contour(): Generates contour plots.

Specialized Visualizations:
 go.Scatter3d(): Generates 3D scatter plots.
 go.Choropleth(): Creates choropleth maps.
 go.FigureWidget(): Enables interactive figures with widgets for live updating.

Customization and Layout:

 plotly.graph_objs.Layout: Allows configuring layout settings for the plot.
 update_layout(): Updates layout settings of the plot.

Interactivity and Animation:

 plotly.express: High-level functions for creating interactive plots easily.
 plotly.graph_objs.Figure: Creates interactive figures.
 plotly.graph_objs.Plot() and plotly.graph_objs.iplot(): Render plotly figures in Jupyter notebooks.

Dash Integration:
 dash_core_components.Graph(): Integrates Plotly graphs into Dash web applications.
 dash_html_components.Div(): Creates HTML div elements for organizing the layout in Dash apps.

Export and Sharing:

 plotly.io.write_image(): Saves figures as images (PNG, JPEG, SVG, etc.).
 plotly.io.show(): Displays plots in Jupyter Notebooks or standalone HTML pages.
 plotly.io.write_html(): Saves figures as standalone HTML files.

Page | 10
Author: Muhammad Omar Akhlaq
 Themes and Styling:
 plotly.io.templates: Provides built-in templates for different plot styles.
 update_traces(): Allows customization of individual traces within a plot.

Dashboards and Web Apps:

 dash: Integrates Plotly plots into web applications using Dash, a Python web framework.

10. NLTK
Corpus and Text Processing:
 nltk.corpus: Module for accessing built-in corpora and lexical resources.
 nltk.word_tokenize(): Tokenizes text into words or sentences.
 nltk.sent_tokenize(): Tokenizes text into sentences.

Basic Text Processing and Analysis:

 nltk.FreqDist(): Generates frequency distributions of words.
 nltk.Text(): Wraps a sequence of tokens for advanced operations like concordance and similar context
search.

Part-of-Speech Tagging:
 nltk.pos_tag(): Assigns parts of speech (POS) tags to words in a text.

Stemming and Lemmatization:

 nltk.PorterStemmer(), nltk.LancasterStemmer(): Implements stemming algorithms.
 nltk.WordNetLemmatizer(): Lemmatizes words based on WordNet.

Named Entity Recognition (NER):

 nltk.ne_chunk(): Labels named entities such as persons, organizations, locations, etc.

Parsing and Syntax:

 nltk.ChartParser(), nltk.RecursiveDescentParser(): Implements parsers for syntactic analysis.
 nltk.ParentedTree(): Represents constituency-based parse trees.

WordNet Interface:
 nltk.WordNet: Interface to access WordNet, a lexical database for English.
 nltk.synsets(): Retrieves synsets (sets of synonyms) from WordNet.

Text Classification:
 nltk.classify: Module containing various classifiers for text classification.
 nltk.NaiveBayesClassifier(), nltk.DecisionTreeClassifier(): Examples of classifiers.

Machine Learning for NLP:

 nltk.classify.scikitlearn: Integrates NLTK classifiers with scikit-learn for machine learning.

Page | 11
Author: Muhammad Omar Akhlaq
 Tokenization and Chunking:
 nltk.chunk: Module for chunking and extracting phrases from sentences.
 nltk.RegexpParser(): Creates chunk parsers using regular expressions.

Sentiment Analysis:
 nltk.sentiment: Module providing sentiment analysis tools and lexicons.

Collocations and Bigrams:

 nltk.collocations: Module for extracting collocations and bigrams from text.
Language Models and Probabilities:
 nltk.probability: Module for implementing probability distributions and frequency estimation.

Page | 12
Author: Muhammad Omar Akhlaq