Top 10 Python Libraries for Data Science
“The goal is to turn data into information and information into insight.”
Carly Fiorina — Former chief executive officer, Hewlett Packard.
There are so many wonderful libraries being added to the Python ecosystem every single day but these are the ones I’ve found the most intuitive and useful (and unsurprisingly the most popular)!
- Numpy - Data Processing
✔ Powerful and complex mathematical operations
✔ Supports multi-dimensional arrays
2. SciPy - Data Processing
✔ Advanced operations on top of Numpy such as integral calculus, linear algebra and probability theory
✔ Supports signal processing
3. Pandas - Data Cleaning & Analysis
✔ Easy to manipulate large amounts of data
✔ Variety of built-in data types e.g. series, dataframes
5. Seaborn - Data Visualisation
✔ Closely integrated with Pandas
✔ Greats visuals
✔ Deals with complex data
6. SciKit-Learn - Data Analysis & Mining
✔ Supports a wide range of algorithms e.g. classification, clustering, dimensionality reduction etc.
✔ Provides plenty of metrics and visuals to determine the accuracy of supervised and unsupervised models
✔ Very intuitive!
7. TensorFlow — Deep Learning
✔ Great for object identification and speech recognition
✔ Models tend to be quite efficient for large-scale systems
✔ Fantastic community support
8. Keras - Deep Learning
✔ Complex models can be created by combining existing neural network models
✔ Great built-in visualisations
✔ Supports most neural network models e.g. convolutional, embedding, fully connected, recurrent etc.
9. PyTorch - Computer Vision & Natural Language Processing (NLP)
✔ Powerful tensor computations
✔ Helps create dynamic computational graphs
10. NLTK - Natural Language Processing (NLP)
✔ Supports both symbolic and statistical NLP
✔ Offers lexical analysis
