Discover packages and tools that have been pivotal in my coding journey as ML engineer. They not only enhance efficiency but also introduce innovative solutions, reshaping how I tackle problems using Python.
In this series, we will explore five, or less, packages from various categories: ML, Data Engineering Pipelines, Frameworks & DL, Visualization, API & Deployment, Developers Tools, and other Packages I Adore.
This installment is centered on Machine Learning packages. Each package comes with a succinct description, its main advantages, and a sample use-case to highlight it’s code design. Where relevant, I’ll also provide alternatives or complimentary packages, giving you a holistic perspective on the tools available.
Machine Learning
1. scikit-learn
Description: A comprehensive library for machine learning algorithms.
Advantage: User-friendly with a consistent API and thorough documentation.
When to use: It’s the package of choice for standard machine learning tasks, including classification, regression, and clustering.
<span>from</span> <span>sklearn.compose</span> <span>import</span> <span>ColumnTransformer</span><span>from</span> <span>sklearn.pipeline</span> <span>import</span> <span>Pipeline</span><span>from</span> <span>sklearn.impute</span> <span>import</span> <span>SimpleImputer</span><span>from</span> <span>sklearn.preprocessing</span> <span>import</span> <span>StandardScaler</span><span>,</span> <span>OneHotEncoder</span><span>from</span> <span>sklearn.feature_extraction.text</span> <span>import</span> <span>TfidfVectorizer</span><span>from</span> <span>sklearn.linear_models</span> <span>import</span> <span>LogisticRegression</span><span>numeric_features</span> <span>=</span> <span>[</span><span>'</span><span>Salary</span><span>'</span><span>]</span><span>numeric_transformer</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[</span><span>(</span><span>'</span><span>imputer</span><span>'</span><span>,</span> <span>SimpleImputer</span><span>(</span><span>strategy</span><span>=</span><span>'</span><span>median</span><span>'</span><span>)),</span><span>(</span><span>'</span><span>scaler</span><span>'</span><span>,</span> <span>StandardScaler</span><span>())])</span><span>text_feature</span> <span>=</span> <span>'</span><span>SelfDescription</span><span>'</span><span>text_transformer</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[</span><span>(</span><span>'</span><span>vectorizer</span><span>'</span><span>,</span> <span>TfidfVectorizer</span><span>(</span><span>stop_words</span><span>=</span><span>"</span><span>english</span><span>"</span><span>))</span><span>])</span><span>categorical_features</span> <span>=</span> <span>[</span><span>'</span><span>Age</span><span>'</span><span>,</span><span>'</span><span>Country</span><span>'</span><span>]</span><span>categorical_transformer</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[</span><span>(</span><span>'</span><span>imputer</span><span>'</span><span>,</span> <span>SimpleImputer</span><span>(</span><span>strategy</span><span>=</span><span>'</span><span>constant</span><span>'</span><span>,</span> <span>fill_value</span><span>=</span><span>'</span><span>missing</span><span>'</span><span>)),</span><span>(</span><span>'</span><span>onehot</span><span>'</span><span>,</span> <span>OneHotEncoder</span><span>(</span><span>handle_unknown</span><span>=</span><span>'</span><span>ignore</span><span>'</span><span>))])</span><span>preprocessor</span> <span>=</span> <span>ColumnTransformer</span><span>(</span><span>transformers</span><span>=</span><span>[</span><span>(</span><span>'</span><span>num</span><span>'</span><span>,</span> <span>numeric_transformer</span><span>,</span> <span>numeric_features</span><span>),</span><span>(</span><span>'</span><span>txt</span><span>'</span><span>,</span> <span>text_transformer</span><span>,</span> <span>text_feature</span><span>),</span><span>(</span><span>'</span><span>cat</span><span>'</span><span>,</span> <span>categorical_transformer</span><span>,</span> <span>categorical_features</span><span>),</span><span>])</span><span>predictor</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[(</span><span>'</span><span>preprocessor</span><span>'</span><span>,</span> <span>preprocessor</span><span>),</span><span>(</span><span>'</span><span>classifier</span><span>'</span><span>,</span> <span>LogisticRegression</span><span>(</span><span>solver</span><span>=</span><span>'</span><span>lbfgs</span><span>'</span><span>))])</span><span># train </span><span>predictor</span><span>.</span><span>fit</span><span>(</span><span>X_train</span><span>,</span> <span>y_train</span><span>)</span><span>#evaluate and predict </span><span>from</span> <span>sklearn.compose</span> <span>import</span> <span>ColumnTransformer</span> <span>from</span> <span>sklearn.pipeline</span> <span>import</span> <span>Pipeline</span> <span>from</span> <span>sklearn.impute</span> <span>import</span> <span>SimpleImputer</span> <span>from</span> <span>sklearn.preprocessing</span> <span>import</span> <span>StandardScaler</span><span>,</span> <span>OneHotEncoder</span> <span>from</span> <span>sklearn.feature_extraction.text</span> <span>import</span> <span>TfidfVectorizer</span> <span>from</span> <span>sklearn.linear_models</span> <span>import</span> <span>LogisticRegression</span> <span>numeric_features</span> <span>=</span> <span>[</span><span>'</span><span>Salary</span><span>'</span><span>]</span> <span>numeric_transformer</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[</span> <span>(</span><span>'</span><span>imputer</span><span>'</span><span>,</span> <span>SimpleImputer</span><span>(</span><span>strategy</span><span>=</span><span>'</span><span>median</span><span>'</span><span>)),</span> <span>(</span><span>'</span><span>scaler</span><span>'</span><span>,</span> <span>StandardScaler</span><span>())])</span> <span>text_feature</span> <span>=</span> <span>'</span><span>SelfDescription</span><span>'</span> <span>text_transformer</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[</span> <span>(</span><span>'</span><span>vectorizer</span><span>'</span><span>,</span> <span>TfidfVectorizer</span><span>(</span><span>stop_words</span><span>=</span><span>"</span><span>english</span><span>"</span><span>))</span> <span>])</span> <span>categorical_features</span> <span>=</span> <span>[</span><span>'</span><span>Age</span><span>'</span><span>,</span><span>'</span><span>Country</span><span>'</span><span>]</span> <span>categorical_transformer</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[</span> <span>(</span><span>'</span><span>imputer</span><span>'</span><span>,</span> <span>SimpleImputer</span><span>(</span><span>strategy</span><span>=</span><span>'</span><span>constant</span><span>'</span><span>,</span> <span>fill_value</span><span>=</span><span>'</span><span>missing</span><span>'</span><span>)),</span> <span>(</span><span>'</span><span>onehot</span><span>'</span><span>,</span> <span>OneHotEncoder</span><span>(</span><span>handle_unknown</span><span>=</span><span>'</span><span>ignore</span><span>'</span><span>))])</span> <span>preprocessor</span> <span>=</span> <span>ColumnTransformer</span><span>(</span> <span>transformers</span><span>=</span><span>[</span> <span>(</span><span>'</span><span>num</span><span>'</span><span>,</span> <span>numeric_transformer</span><span>,</span> <span>numeric_features</span><span>),</span> <span>(</span><span>'</span><span>txt</span><span>'</span><span>,</span> <span>text_transformer</span><span>,</span> <span>text_feature</span><span>),</span> <span>(</span><span>'</span><span>cat</span><span>'</span><span>,</span> <span>categorical_transformer</span><span>,</span> <span>categorical_features</span><span>),</span> <span>])</span> <span>predictor</span> <span>=</span> <span>Pipeline</span><span>(</span><span>steps</span><span>=</span><span>[(</span><span>'</span><span>preprocessor</span><span>'</span><span>,</span> <span>preprocessor</span><span>),</span> <span>(</span><span>'</span><span>classifier</span><span>'</span><span>,</span> <span>LogisticRegression</span><span>(</span><span>solver</span><span>=</span><span>'</span><span>lbfgs</span><span>'</span><span>))])</span> <span># train </span><span>predictor</span><span>.</span><span>fit</span><span>(</span><span>X_train</span><span>,</span> <span>y_train</span><span>)</span> <span>#evaluate and predict </span>from sklearn.compose import ColumnTransformer from sklearn.pipeline import Pipeline from sklearn.impute import SimpleImputer from sklearn.preprocessing import StandardScaler, OneHotEncoder from sklearn.feature_extraction.text import TfidfVectorizer from sklearn.linear_models import LogisticRegression numeric_features = ['Salary'] numeric_transformer = Pipeline(steps=[ ('imputer', SimpleImputer(strategy='median')), ('scaler', StandardScaler())]) text_feature = 'SelfDescription' text_transformer = Pipeline(steps=[ ('vectorizer', TfidfVectorizer(stop_words="english")) ]) categorical_features = ['Age','Country'] categorical_transformer = Pipeline(steps=[ ('imputer', SimpleImputer(strategy='constant', fill_value='missing')), ('onehot', OneHotEncoder(handle_unknown='ignore'))]) preprocessor = ColumnTransformer( transformers=[ ('num', numeric_transformer, numeric_features), ('txt', text_transformer, text_feature), ('cat', categorical_transformer, categorical_features), ]) predictor = Pipeline(steps=[('preprocessor', preprocessor), ('classifier', LogisticRegression(solver='lbfgs'))]) # train predictor.fit(X_train, y_train) #evaluate and predict
Enter fullscreen mode Exit fullscreen mode
river complements scikit-learn by offering tools specifically designed for online learning, ideal for scenarios where data is streaming in real-time. While scikit-learn is optimized for batch learning, river provides a solution for incrementally updating models with new data points as they arrive.
skorch seamlessly integrates the deep learning capabilities of PyTorch into the scikit-learn ecosystem. It allows developers to use PyTorch-based neural networks as if they were scikit-learn estimators, making it easier to incorporate deep learning models into workflows that already leverage scikit-learn tools, such as grid search and pipelines.
2. PyMC
Description: Specialized in Bayesian modeling and probabilistic machine learning.
Advantage: Equips you with the tools to define probabilistic models in code.
When to use: Ideal for white-box ML using probabilistic programming.
<span>import</span> <span>pymc</span> <span>as</span> <span>pm</span><span>import</span> <span>xarray</span> <span>as</span> <span>xr</span><span>with</span> <span>pm</span><span>.</span><span>Model</span><span>()</span> <span>as</span> <span>model</span><span>:</span><span># Priors </span> <span>alpha</span> <span>=</span> <span>pm</span><span>.</span><span>Normal</span><span>(</span><span>'</span><span>alpha</span><span>'</span><span>,</span> <span>mu</span><span>=</span><span>0</span><span>,</span> <span>sd</span><span>=</span><span>10</span><span>)</span><span>beta</span> <span>=</span> <span>pm</span><span>.</span><span>Normal</span><span>(</span><span>'</span><span>beta</span><span>'</span><span>,</span> <span>mu</span><span>=</span><span>0</span><span>,</span> <span>sd</span><span>=</span><span>10</span><span>,</span> <span>shape</span><span>=</span><span>X_train</span><span>.</span><span>shape</span><span>[</span><span>1</span><span>])</span><span># Linear combination </span> <span>mu</span> <span>=</span> <span>alpha</span> <span>+</span> <span>xr</span><span>.</span><span>dot</span><span>(</span><span>X_train</span><span>,</span> <span>beta</span><span>)</span><span># Logistic link function </span> <span>p</span> <span>=</span> <span>pm</span><span>.</span><span>invlogit</span><span>(</span><span>mu</span><span>)</span><span># Likelihood </span> <span>y_obs</span> <span>=</span> <span>pm</span><span>.</span><span>Bernoulli</span><span>(</span><span>'</span><span>y_obs</span><span>'</span><span>,</span> <span>p</span><span>=</span><span>p</span><span>,</span> <span>observed</span><span>=</span><span>y_train</span><span>)</span><span># Sample/train </span> <span>trace</span> <span>=</span> <span>pm</span><span>.</span><span>sample</span><span>(</span><span>3000</span><span>)</span><span># Evaluation with posterior predictive checks # Prediction by drawing samples from the posterior predictive distribution </span><span>import</span> <span>pymc</span> <span>as</span> <span>pm</span> <span>import</span> <span>xarray</span> <span>as</span> <span>xr</span> <span>with</span> <span>pm</span><span>.</span><span>Model</span><span>()</span> <span>as</span> <span>model</span><span>:</span> <span># Priors </span> <span>alpha</span> <span>=</span> <span>pm</span><span>.</span><span>Normal</span><span>(</span><span>'</span><span>alpha</span><span>'</span><span>,</span> <span>mu</span><span>=</span><span>0</span><span>,</span> <span>sd</span><span>=</span><span>10</span><span>)</span> <span>beta</span> <span>=</span> <span>pm</span><span>.</span><span>Normal</span><span>(</span><span>'</span><span>beta</span><span>'</span><span>,</span> <span>mu</span><span>=</span><span>0</span><span>,</span> <span>sd</span><span>=</span><span>10</span><span>,</span> <span>shape</span><span>=</span><span>X_train</span><span>.</span><span>shape</span><span>[</span><span>1</span><span>])</span> <span># Linear combination </span> <span>mu</span> <span>=</span> <span>alpha</span> <span>+</span> <span>xr</span><span>.</span><span>dot</span><span>(</span><span>X_train</span><span>,</span> <span>beta</span><span>)</span> <span># Logistic link function </span> <span>p</span> <span>=</span> <span>pm</span><span>.</span><span>invlogit</span><span>(</span><span>mu</span><span>)</span> <span># Likelihood </span> <span>y_obs</span> <span>=</span> <span>pm</span><span>.</span><span>Bernoulli</span><span>(</span><span>'</span><span>y_obs</span><span>'</span><span>,</span> <span>p</span><span>=</span><span>p</span><span>,</span> <span>observed</span><span>=</span><span>y_train</span><span>)</span> <span># Sample/train </span> <span>trace</span> <span>=</span> <span>pm</span><span>.</span><span>sample</span><span>(</span><span>3000</span><span>)</span> <span># Evaluation with posterior predictive checks # Prediction by drawing samples from the posterior predictive distribution </span>import pymc as pm import xarray as xr with pm.Model() as model: # Priors alpha = pm.Normal('alpha', mu=0, sd=10) beta = pm.Normal('beta', mu=0, sd=10, shape=X_train.shape[1]) # Linear combination mu = alpha + xr.dot(X_train, beta) # Logistic link function p = pm.invlogit(mu) # Likelihood y_obs = pm.Bernoulli('y_obs', p=p, observed=y_train) # Sample/train trace = pm.sample(3000) # Evaluation with posterior predictive checks # Prediction by drawing samples from the posterior predictive distribution
Enter fullscreen mode Exit fullscreen mode
While Stan offers its own modeling language and provides MCMC sampling, and Edward integrates with TensorFlow/Keras to offer Variational Inference, PyMC stands out for its ease of use within the Python environment, user-friendly API, and active community.
3. darts
Description: My preferred package for time series forecasting and anomaly detection.
Advantage: It offers comprehensive tools for time series analysis and a unified interface for various forecasting models.
When to use: Essential when dealing with time series data and you need forecasting, anomaly detection, or other analyses using classical, deep learning, prophet models, and beyond.
<span>from</span> <span>darts.models</span> <span>import</span> <span>RNNModel</span><span>model_config</span> <span>=</span> <span>{</span><span>"</span><span>model_name</span><span>"</span><span>:</span> <span>"</span><span>Sales_LSTM</span><span>"</span><span>,</span><span>"</span><span>hidden_dim</span><span>"</span><span>:</span> <span>20</span><span>,</span><span>"</span><span>dropout</span><span>"</span><span>:</span> <span>0</span><span>,</span><span>"</span><span>batch_size</span><span>"</span><span>:</span> <span>16</span><span>,</span><span>"</span><span>n_epochs</span><span>"</span><span>:</span> <span>200</span><span>,</span><span>"</span><span>random_state</span><span>"</span><span>:</span> <span>42</span><span>,</span><span>"</span><span>training_length</span><span>"</span><span>:</span> <span>20</span><span>,</span><span>"</span><span>input_chunk_length</span><span>"</span><span>:</span> <span>14</span><span>,</span><span>"</span><span>force_reset</span><span>"</span><span>:</span> <span>True</span><span>,</span><span>"</span><span>save_checkpoints</span><span>"</span><span>:</span> <span>True</span><span>,</span><span>}</span><span>model</span> <span>=</span> <span>RNNModel</span><span>(</span><span>model</span><span>=</span><span>"</span><span>LSTM</span><span>"</span><span>,</span><span>optimizer_kwargs</span><span>=</span><span>{</span><span>"</span><span>lr</span><span>"</span><span>:</span> <span>1e-3</span><span>},</span><span>**</span><span>model_config</span><span>)</span><span># train </span><span>model</span><span>.</span><span>fit</span><span>(</span><span>TimeSeriesData</span><span>)</span><span># forecast next 3 </span><span>forecast</span> <span>=</span> <span>model</span><span>.</span><span>predict</span><span>(</span><span>3</span><span>)</span><span>from</span> <span>darts.models</span> <span>import</span> <span>RNNModel</span> <span>model_config</span> <span>=</span> <span>{</span> <span>"</span><span>model_name</span><span>"</span><span>:</span> <span>"</span><span>Sales_LSTM</span><span>"</span><span>,</span> <span>"</span><span>hidden_dim</span><span>"</span><span>:</span> <span>20</span><span>,</span> <span>"</span><span>dropout</span><span>"</span><span>:</span> <span>0</span><span>,</span> <span>"</span><span>batch_size</span><span>"</span><span>:</span> <span>16</span><span>,</span> <span>"</span><span>n_epochs</span><span>"</span><span>:</span> <span>200</span><span>,</span> <span>"</span><span>random_state</span><span>"</span><span>:</span> <span>42</span><span>,</span> <span>"</span><span>training_length</span><span>"</span><span>:</span> <span>20</span><span>,</span> <span>"</span><span>input_chunk_length</span><span>"</span><span>:</span> <span>14</span><span>,</span> <span>"</span><span>force_reset</span><span>"</span><span>:</span> <span>True</span><span>,</span> <span>"</span><span>save_checkpoints</span><span>"</span><span>:</span> <span>True</span><span>,</span> <span>}</span> <span>model</span> <span>=</span> <span>RNNModel</span><span>(</span> <span>model</span><span>=</span><span>"</span><span>LSTM</span><span>"</span><span>,</span> <span>optimizer_kwargs</span><span>=</span><span>{</span><span>"</span><span>lr</span><span>"</span><span>:</span> <span>1e-3</span><span>},</span> <span>**</span><span>model_config</span> <span>)</span> <span># train </span><span>model</span><span>.</span><span>fit</span><span>(</span><span>TimeSeriesData</span><span>)</span> <span># forecast next 3 </span><span>forecast</span> <span>=</span> <span>model</span><span>.</span><span>predict</span><span>(</span><span>3</span><span>)</span>from darts.models import RNNModel model_config = { "model_name": "Sales_LSTM", "hidden_dim": 20, "dropout": 0, "batch_size": 16, "n_epochs": 200, "random_state": 42, "training_length": 20, "input_chunk_length": 14, "force_reset": True, "save_checkpoints": True, } model = RNNModel( model="LSTM", optimizer_kwargs={"lr": 1e-3}, **model_config ) # train model.fit(TimeSeriesData) # forecast next 3 forecast = model.predict(3)
Enter fullscreen mode Exit fullscreen mode
Alternatives: Merlion and kats
While Merlion and Kats offer their own sets of capabilities in time series analysis, Darts shines as a comprehensive choice for time series forecasting and processing, catering to a wide range of requirements with its extensive toolkit. Both Merlion and Kats can serve as potential alternatives, but Darts’ holistic offerings make it a standout choice for me.
4. FLAML
Description: A swift and efficient automated machine learning library.
Advantage: Achieve optimal ML results with minimal coding and time investment.
When to use: Perfect when you desire swift outcomes without the intricacies of model fine-tuning.
<span>from</span> <span>flaml</span> <span>import</span> <span>AutoML</span><span>automl</span> <span>=</span> <span>AutoML</span><span>()</span><span>automl_config</span> <span>=</span> <span>{</span><span>"</span><span>time_budget</span><span>"</span><span>:</span> <span>120</span><span>,</span> <span># time in seconds </span> <span>"</span><span>metric</span><span>"</span><span>:</span> <span>'</span><span>accuracy</span><span>'</span><span>,</span><span>"</span><span>task</span><span>"</span><span>:</span> <span>'</span><span>classification</span><span>'</span><span>,</span><span>"</span><span>estimator_list</span><span>"</span><span>:</span> <span>[</span><span>'</span><span>lgbm</span><span>'</span><span>,</span> <span>'</span><span>xgboost</span><span>'</span><span>,</span> <span>'</span><span>catboost</span><span>'</span><span>,</span> <span>'</span><span>extra_tree</span><span>'</span><span>,],</span><span>"</span><span>seed</span><span>"</span><span>:</span> <span>42</span><span>,</span><span>"</span><span>log_file_name</span><span>"</span><span>:</span> <span>"</span><span>churn.log</span><span>"</span><span>,</span><span>"</span><span>log_training_metric</span><span>"</span><span>:</span> <span>True</span><span>,</span><span>}</span><span># train </span><span>automl</span><span>.</span><span>fit</span><span>(</span><span>X_train</span><span>,</span> <span>y_train</span><span>,</span> <span>**</span><span>automl_config</span><span>)</span><span># evaluate and predict </span><span>from</span> <span>flaml</span> <span>import</span> <span>AutoML</span> <span>automl</span> <span>=</span> <span>AutoML</span><span>()</span> <span>automl_config</span> <span>=</span> <span>{</span> <span>"</span><span>time_budget</span><span>"</span><span>:</span> <span>120</span><span>,</span> <span># time in seconds </span> <span>"</span><span>metric</span><span>"</span><span>:</span> <span>'</span><span>accuracy</span><span>'</span><span>,</span> <span>"</span><span>task</span><span>"</span><span>:</span> <span>'</span><span>classification</span><span>'</span><span>,</span> <span>"</span><span>estimator_list</span><span>"</span><span>:</span> <span>[</span><span>'</span><span>lgbm</span><span>'</span><span>,</span> <span>'</span><span>xgboost</span><span>'</span><span>,</span> <span>'</span><span>catboost</span><span>'</span><span>,</span> <span>'</span><span>extra_tree</span><span>'</span><span>,],</span> <span>"</span><span>seed</span><span>"</span><span>:</span> <span>42</span><span>,</span> <span>"</span><span>log_file_name</span><span>"</span><span>:</span> <span>"</span><span>churn.log</span><span>"</span><span>,</span> <span>"</span><span>log_training_metric</span><span>"</span><span>:</span> <span>True</span><span>,</span> <span>}</span> <span># train </span><span>automl</span><span>.</span><span>fit</span><span>(</span><span>X_train</span><span>,</span> <span>y_train</span><span>,</span> <span>**</span><span>automl_config</span><span>)</span> <span># evaluate and predict </span>from flaml import AutoML automl = AutoML() automl_config = { "time_budget": 120, # time in seconds "metric": 'accuracy', "task": 'classification', "estimator_list": ['lgbm', 'xgboost', 'catboost', 'extra_tree',], "seed": 42, "log_file_name": "churn.log", "log_training_metric": True, } # train automl.fit(X_train, y_train, **automl_config) # evaluate and predict
Enter fullscreen mode Exit fullscreen mode
Complimentary: AutoGluon and mljar-supervised
FLAML specializes in automating machine learning tasks for tabular data. In contrast, AutoGluon amplifies the automation game by accommodating a wider spectrum, including text, images, and multi-modal data, making it a more versatile toolkit. Meanwhile, mljar-supervised extends FLAML by adding model explanation, ensemble, and visualization, , presenting itself as a viable alternative with comparable capabilities.
5. CVXPY
Description: The go-to library for convex optimization.
Advantage: It provides an intuitive method to define and solve convex optimisation problems.
When to use: Essential for solving optimisation challenges across domains like finance, control, signal processing, and more.
<span>'''</span><span> Task: Operations Research PYIKEA wants to maximize its profit of selling armchair, wingchair, and Lovet-table. The profit of selling armchair is 150 DKK, wingchair 100 DKK, and Lovet-table 250 DKK. It takes: 15 planks of wood and 5 hours of labour to make one armchair 12 planks of wood and 2 hours of labour to make one wingchair 18 planks of wood and 8 hours of labour to make one Lovet-table The store needs at least 4 of each chair, and a table. The total amount of woods pieces in storage is 450 and labour budget is of 120 hours only. What combination of chairs and table(s) yield maximum profit? </span><span>'''</span><span>import</span> <span>cvxpy</span> <span>as</span> <span>cp</span><span># Variables </span><span>A</span> <span>=</span> <span>cp</span><span>.</span><span>Variable</span><span>(</span><span>integer</span><span>=</span><span>True</span><span>,</span> <span>name</span><span>=</span><span>"</span><span>Armchair</span><span>"</span><span>)</span><span>W</span> <span>=</span> <span>cp</span><span>.</span><span>Variable</span><span>(</span><span>integer</span><span>=</span><span>True</span><span>,</span> <span>name</span><span>=</span><span>"</span><span>Wingchair</span><span>"</span><span>)</span><span>L</span> <span>=</span> <span>cp</span><span>.</span><span>Variable</span><span>(</span><span>integer</span><span>=</span><span>True</span><span>,</span> <span>name</span><span>=</span><span>"</span><span>Lovet-table</span><span>"</span><span>)</span><span># Objective </span><span>profit</span> <span>=</span> <span>150</span><span>*</span><span>A</span> <span>+</span> <span>100</span><span>*</span><span>W</span> <span>+</span> <span>250</span><span>*</span><span>L</span><span>objective</span> <span>=</span> <span>cp</span><span>.</span><span>Maximize</span><span>(</span><span>profit</span><span>)</span><span># Constraints </span><span>constraints</span> <span>=</span> <span>[</span><span>15</span><span>*</span><span>A</span> <span>+</span> <span>12</span><span>*</span><span>W</span> <span>+</span> <span>18</span><span>*</span><span>L</span> <span><=</span> <span>450</span><span>,</span><span>5</span><span>*</span><span>A</span> <span>+</span> <span>2</span><span>*</span><span>W</span> <span>+</span> <span>8</span><span>*</span><span>L</span> <span><=</span> <span>120</span><span>,</span><span>A</span> <span>>=</span> <span>4</span><span>,</span><span>W</span> <span>>=</span> <span>4</span><span>,</span><span>L</span> <span>>=</span> <span>1</span><span>]</span><span># Problem to Solve </span><span>problem</span> <span>=</span> <span>cp</span><span>.</span><span>Problem</span><span>(</span><span>objective</span><span>,</span> <span>constraints</span><span>=</span><span>constraints</span><span>)</span><span>result</span> <span>=</span> <span>problem</span><span>.</span><span>solve</span><span>()</span><span>'''</span><span> Task: Operations Research PYIKEA wants to maximize its profit of selling armchair, wingchair, and Lovet-table. The profit of selling armchair is 150 DKK, wingchair 100 DKK, and Lovet-table 250 DKK. It takes: 15 planks of wood and 5 hours of labour to make one armchair 12 planks of wood and 2 hours of labour to make one wingchair 18 planks of wood and 8 hours of labour to make one Lovet-table The store needs at least 4 of each chair, and a table. The total amount of woods pieces in storage is 450 and labour budget is of 120 hours only. What combination of chairs and table(s) yield maximum profit? </span><span>'''</span> <span>import</span> <span>cvxpy</span> <span>as</span> <span>cp</span> <span># Variables </span><span>A</span> <span>=</span> <span>cp</span><span>.</span><span>Variable</span><span>(</span><span>integer</span><span>=</span><span>True</span><span>,</span> <span>name</span><span>=</span><span>"</span><span>Armchair</span><span>"</span><span>)</span> <span>W</span> <span>=</span> <span>cp</span><span>.</span><span>Variable</span><span>(</span><span>integer</span><span>=</span><span>True</span><span>,</span> <span>name</span><span>=</span><span>"</span><span>Wingchair</span><span>"</span><span>)</span> <span>L</span> <span>=</span> <span>cp</span><span>.</span><span>Variable</span><span>(</span><span>integer</span><span>=</span><span>True</span><span>,</span> <span>name</span><span>=</span><span>"</span><span>Lovet-table</span><span>"</span><span>)</span> <span># Objective </span><span>profit</span> <span>=</span> <span>150</span><span>*</span><span>A</span> <span>+</span> <span>100</span><span>*</span><span>W</span> <span>+</span> <span>250</span><span>*</span><span>L</span> <span>objective</span> <span>=</span> <span>cp</span><span>.</span><span>Maximize</span><span>(</span><span>profit</span><span>)</span> <span># Constraints </span><span>constraints</span> <span>=</span> <span>[</span><span>15</span><span>*</span><span>A</span> <span>+</span> <span>12</span><span>*</span><span>W</span> <span>+</span> <span>18</span><span>*</span><span>L</span> <span><=</span> <span>450</span><span>,</span> <span>5</span><span>*</span><span>A</span> <span>+</span> <span>2</span><span>*</span><span>W</span> <span>+</span> <span>8</span><span>*</span><span>L</span> <span><=</span> <span>120</span><span>,</span> <span>A</span> <span>>=</span> <span>4</span><span>,</span> <span>W</span> <span>>=</span> <span>4</span><span>,</span> <span>L</span> <span>>=</span> <span>1</span> <span>]</span> <span># Problem to Solve </span><span>problem</span> <span>=</span> <span>cp</span><span>.</span><span>Problem</span><span>(</span><span>objective</span><span>,</span> <span>constraints</span><span>=</span><span>constraints</span><span>)</span> <span>result</span> <span>=</span> <span>problem</span><span>.</span><span>solve</span><span>()</span>''' Task: Operations Research PYIKEA wants to maximize its profit of selling armchair, wingchair, and Lovet-table. The profit of selling armchair is 150 DKK, wingchair 100 DKK, and Lovet-table 250 DKK. It takes: 15 planks of wood and 5 hours of labour to make one armchair 12 planks of wood and 2 hours of labour to make one wingchair 18 planks of wood and 8 hours of labour to make one Lovet-table The store needs at least 4 of each chair, and a table. The total amount of woods pieces in storage is 450 and labour budget is of 120 hours only. What combination of chairs and table(s) yield maximum profit? ''' import cvxpy as cp # Variables A = cp.Variable(integer=True, name="Armchair") W = cp.Variable(integer=True, name="Wingchair") L = cp.Variable(integer=True, name="Lovet-table") # Objective profit = 150*A + 100*W + 250*L objective = cp.Maximize(profit) # Constraints constraints = [15*A + 12*W + 18*L <= 450, 5*A + 2*W + 8*L <= 120, A >= 4, W >= 4, L >= 1 ] # Problem to Solve problem = cp.Problem(objective, constraints=constraints) result = problem.solve()
Enter fullscreen mode Exit fullscreen mode
Alternative: pyomo
Pyomo and cvxpy are both my interstellar choices for optimisation in Python, each with its own set of strengths. While cvxpy excels with its intuitive approach to convex problems, Pyomo flaunts versatility in tackling a variety of optimisation challenges, be it linear, nonlinear, or mixed-integer. Essentially, picking between the two boils down to the mood I am on of the day!
We navigated through my favourite Python packages that redefine machine learning workflows. I picked scikit-learn for general ML tasks, complemented by tools like river and skorch. PyMC guided us through the intricacies of Bayesian modeling with alternatives like Stan and Edward. darts emerged as a comprehensive choice for time series analysis, though Merlion and kats offer their unique capabilities.
For rapid results, FLAML streamlines automated ML, with AutoGluon and mljar-supervised expanding on similar terrains. Lastly, CVXPY showcased its prowess in optimization, with pyomo as another contender. These packages, collectively, illuminate the expansive and evolving landscape of Python-based machine learning that can elevate your skills, as they did mine.
Stay tuned for the next segment on Data Engineering Pipelines, featuring Dagster, Apache Airflow, Prefect and Argo.
Until then, stay curious and keep on coding.
原文链接：Elevate Your Python Skills: Machine Learning Packages That Transformed My Journey as ML Engineer
© 版权声明
THE END
暂无评论内容