Data Science Keys to Open Up OpenNASA Datasets
1 like269 views
The document discusses the utilization of Natural Language Processing (NLP) techniques on NASA datasets, emphasizing data sources, metadata, and methods for data analysis, including topic modeling and clustering approaches. It outlines the importance of preprocessing text data and presents various tools and libraries available for analyzing such datasets. Key takeaways include the potential of NLP in understanding structured and unstructured text as well as the utility of open government data for public access and insight.
Data Science Keys to Open Up OpenNASA Datasets
- 1. PyData New York 2017 Noemi Derzsy Data Science Keys to Open Up OpenNASA Datasets @NoemiDerzsy
- 2. PyData New York 2017 Open NASA Platform • Data at NASA: 32 089 • Data at other government agencies: ~185 000 • NASA code repositories: 356 • NASA APIs: 51 Continuously growing… https://open.nasa.gov/explore/datanauts/ @NoemiDerzsy
- 3. PyData New York 2017 @NoemiDerzsy
- 4. PyData New York 2017 § Datasets: 32089 § Some of the data sets: - Mars Rover sound data - Hubble Telescope image collection - NASA patents - Picture of the Day of Earth - etc. http://data.nasa.gov/data.json Open NASA Metadata Metadata information: • id • type • accessLevel • accrualPeriodicity • bureauCode • contactPoint • title • description • distribution • identifier • Issued • keyword • language • modified • programCode • theme • license • location (HTML link) • Etc. Format:JSON @NoemiDerzsy
- 5. PyData New York 2017 Which of these features is “best” to tie together the data? How do we label groupings in a meaningful manner? How many groups/how to arrange/visualize them? Are Descriptions, Keywords representative of the content? @NoemiDerzsy
- 6. PyData New York 2017 Natural Language Processing (NLP) Python Libraries • NLTK • TextBlob • spaCy • gensim • Stanford CoreNLP Jupyter Notebooks: https://github.com/nderzsy/NASADatanauts @NoemiDerzsy
- 7. PyData New York 2017 Word Clouds in Description Word clouds of the over 32,000 open NASA datasets and their descriptions @NoemiDerzsy
- 8. PyData New York 2017 Word clouds of the over 32,000 open NASA datasets and their titles Word Clouds in Title @NoemiDerzsy
- 9. PyData New York 2017 Word Clouds in Keywords Word clouds of the over 32,000 open NASA datasets and their keywords @NoemiDerzsy
- 10. PyData New York 2017 How to Obtain Customized Word Clouds? o get stencil (shape of your choice) o get text o https://github.com/amueller/word_cloud @NoemiDerzsy
- 11. PyData New York 2017 Text Preprocessing, Cleaning • treat “Data” and “data” as identical words: covert all words to lowercase lower() • remove special characters, codes, numbers: regular expressions • check for misspelling • stop words: this, and, for, where, etc. • “system” vs. “systems”: lemmatize • “compute”, “computer”, “computation” -> “comput”: stem • tokenize: break down text to smallest parts (words) • POS tagging • dimensionality reduction (PCA) @NoemiDerzsy
- 12. PyData New York 2017 Stemming • from nltk.stem.api import StemmerI • from nltk.stem.regexp import RegexpStemmer • from nltk.stem.lancaster import LancasterStemmer • from nltk.stem.isri import ISRIStemmer • from nltk.stem.porter import PorterStemmer • from nltk.stem.snowball import SnowballStemmer • from nltk.stem.wordnet import WordNetLemmatizer • from nltk.stem.rslp import RSLPStemmer • Lemmatization: similar to stemming, but with stems being valid words • nltk.WordNetLemmatizer() Lemmatization @NoemiDerzsy
- 13. PyData New York 2017 Term Frequency Keywords Description Title • the number of times a word occurs in text corpus @NoemiDerzsy
- 14. PyData New York 2017 TF-IDF • term frequency – inverse document frequency • measures term frequency / document frequency Top 25 Highest TF-IDF Words in Description Top 25 Highest MeanTF-IDF Words in Description @NoemiDerzsy
- 15. PyData New York 2017 Description TF-IDF and Keywords Top 10 keywords: 1. EARTH SCIENCE: 14387 2. OCEANS: 10034 3. PROJECT: 7464 4. OCEAN OPTICS: 7325 5. ATMOSPHERE: 7324 6. OCEAN COLOR: 7271 7. COMPLETED: 6453 8. ATMOSPHERIC WATER VAPOR: 3143 9. LAND SURFACE: 2721 10.BIOSPHERE: 2450 Top 25 Keywords @NoemiDerzsy
- 16. PyData New York 2017 Word Co-Occurrence § Co-occurrence matrix of top most frequently co-occurring terms Top 10 Word Co-Occurrence Matrix @NoemiDerzsy
- 17. PyData New York 2017 Word Co-Occurrence Top 20 Word Co-Occurrence Matrix @NoemiDerzsy
- 18. PyData New York 2017 Top 50 Word Co-Occurrence Matrix @NoemiDerzsy
- 19. PyData New York 2017 seaborn.clustermap Top 20 Word Co-Occurrence Matrix Discovering Structure in Heatmap Data standardization @NoemiDerzsy
- 20. PyData New York 2017 Are features pulled from text (such as title, description fields) and/or human supplied-keywords descriptive of the content? Topic Modeling… @NoemiDerzsy
- 21. PyData New York 2017 What is Topic Modeling? An efficient way to make sense of large volume of texts. Identify topics within text corpus. Categorize documents into topics. Associate words with topics. Who uses it? Search engines, for marketing purpose, etc. @NoemiDerzsy
- 22. PyData New York 2017 Latent Dirichlet Allocation (LDA) q several techniques, but LDA is the most common q Bayesian inference model that associates each document with a probability distribution over topics q topics are probability distributions over words (probability of the word being generated from that topic for that document) q clusters words into topics q clusters documents into mixture of topics q scales well with growing corpus q before running LDA algorithm,we have to specify the number of topics: how to choose beforehand the optimal number of topics? @NoemiDerzsy
- 23. PyData New York 2017 Topic Model Evaluation: Topic Coherence Q: How to select the top topics? A: Calculate the UMass topic coherence for each topic. Algorithm from Mimno, Wallach, Talley, Leenders, McCallum: Optimizing Semantic Coherence in Topic Models, CEMNLP 2011. Coherence = ! score(𝑤), 𝑤,) )., pairwise scores on the words used to describe the topic. s𝑐𝑜𝑟𝑒34566 78,79 = log 𝐷 𝑤), 𝑤, + 1 𝐷(𝑤)) D(wi)D(wi) as the count of documents containing the word wiwi, D(wi,wj)D(wi,wj) the count of documents containing both words wiwi and wjwj, and DD the total number or documents in the corpus. @NoemiDerzsy
- 24. PyData New York 2017 openNASA Topics of Highest Coherence @NoemiDerzsy
- 25. PyData New York 2017 Keywords for Topics • selected keywords with their most frequently occurring terms @NoemiDerzsy
- 26. PyData New York 2017 Other Clustering Method: K-Means • using TF-IDF, the document vectors are put through a K-Means clustering algorithm which computes the Euclidean distances amongst these documents and clusters nearby documents together • the algorithm generates cluster tags, known as cluster centers which represent the documents within these clusters • K-means distance: • Euclidean • Cosine • Fuzzy • Accuracy comparison: • silhouette analysis can be used to study the separation distance between the resulting clusters; can be used to determine the optimal number of clusters @NoemiDerzsy
- 27. PyData New York 2017 K-Means Clustering § Top 10 terms per cluster: Cluster 0 seawifs local collect km mission data orbview seastar broadcast noon § Similar clusters with top words as found using LDA Cluster 1 data project system use soil high contain phase set gt Cluster 2 modis terra aqua earth global pass south north equator eos Cluster 3 band oct color adeos czcs nominal sense spacecraft agency thermal Cluster 4 product data level version file aquarius set daily ml standard @NoemiDerzsy
- 28. PyData New York 2017 K-Means Clustering § cluster size distribution for k = 5 Cluster 2: 19031 Cluster 0: 5805 Cluster 1: 4481 Cluster 3: 1968 Cluster 4: 804 § silhouette score improves by increasing the number of clusters, best performance for K > 100 § analysis of cluster size distributions with varying K can reveal additional information 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0 50 100 150 200 250 300 350 SilhouetteScore K Number of Clusters Silhouette Score for K-Means Clustering openNASA Datasets @NoemiDerzsy
- 29. PyData New York 2017 Text Classification § Naïve Bayes probabilistic model • Multinomial model • data follows multinomial distribution • each feature value is a count (word co-occurrence, weight, tf-idf, etc.) • Take into account word importance • Bernoulli model • data follows a multivariate Bernoulli distribution • bag of words (count word occurrence) • each feature is a binary feature (word in text? True/False) • ignores word significance § KNN (K-nearest neighbor) classifier § Decision trees § Support Vector Machine (SVM) @NoemiDerzsy
- 30. PyData New York 2017 Network Science: Bipartite Network § Term in Description – Keyword: - connect 2 words in description if they appear under same Keyword - connect 2 keywords if they share common words in Description § All bipartite network projection possibilities: - description – keyword - description – title - title – keyword § Type of interaction: - link direction: directed/undirected - link weight: how strong the connection (co-occurrence matrix) @NoemiDerzsy
- 31. PyData New York 2017 Structural Network Properties § degree distribution: reveals connectivity pattern § do we have hubs (nodes with significantly higher number of connections)? § average degree (how interconnected the elements are) § clusters, community detection (subgroups with elements more densely connected among each other) @NoemiDerzsy
- 32. PyData New York 2017 Terms that occur > 1% of the documents (in more than 320 data set descriptions) • Network size: # of nodes (words) = 1015 • Average degree: 890.319 (densely connected terms) • Components: 1 connected component • Modularity analysis: 2 clusters (pink, violet) @NoemiDerzsy
- 33. PyData New York 2017 Top 100 most frequent words in Description • size/color hue ~ occurrence frequency • outer circle: top 10 most frequent words Nodes with the highest connectivity are not all the same as most frequently co-occurring @NoemiDerzsy
- 34. PyData New York 2017 Network structures: powerful tool for capturing additional information about interconnected systems! @NoemiDerzsy
- 35. PyData New York 2017 What are the important connections between NASA datasets and other important datasets outside of NASA in the US government? @NoemiDerzsy
- 36. PyData New York 2017 Other Open Government Dataset Collections http://data.nasa.gov/data.json http://www.epa.gov/data.json http://data.gov http://www.nsf.gov/data.json http://usda.gov/data.json http://data.noaa.gov/data.json http://www.commerce.gov/data.json http://nist.gov/data.json http://www.defense.gov/data.json http://www2.ed.gov/data.json http://www.dol.gov/data.json http://www.state.gov/data.json http://www.dot.gov/data.json http://www.energy.gov/data.json http://nrel.gov/data.json http://healthdata.gov/data.json http://www.hud.gov/data.json http://www.doi.gov/data.json http://www.justice.gov/data.json http://www.archives.gov/data.json http://www.nrc.gov/data.json http://www.nsf.gov/data.json http://www.opm.gov/data.json https://www.sba.gov/sites/default/files/data.json http://www.ssa.gov/data.json http://www.consumerfinance.gov/data.json http://www.fhfa.gov/data.json http://www.imls.gov/data.json http://data.mcc.gov/raw/index.json http://www.nitrd.gov/data.json http://www.ntsb.gov/data.json http://www.sec.gov/data.json https://open.whitehouse.gov/data.json http://treasury.gov/data.json http://www.usaid.gov/data.json http://www.gsa.gov/data.json https://www.economist.com/news/international/21678833-open-data-revolution-has-not-lived-up-expectations-it-only-getting Open Government Data: Out of the Box The Economist @NoemiDerzsy
- 37. PyData New York 2017 pyNASA and pyOpenGov Libraries § Python library that loads all the open NASA or other government metadata collection at once pyNASA https://github.com/bmtgoncalves/pyNASA pyOpenGov https://github.com/nderzsy/pyOpenGov How to install: >> pip install pyNASA >> pip install pyOpenGov @NoemiDerzsy
- 38. PyData New York 2017 Takeaways § NLP enables understanding of structured and unstructured text § Topic modeling useful tool for understanding topics in large text corpus, documents § Topic models efficient for evaluating the accuracy of human-supplied descriptions, keywords § Tedious preprocessing a must before modeling (stop words, lemmatization, special characters, etc.) § Network (graph) structure can reveal more information, term/topic associations § Network projections allow to answer additional questions § Open government data enables citizens in understanding topics, areas of focus @NoemiDerzsy
- 39. Contact GitHub: https://github.com/nderzsy/NASADatanauts Twitter: @NoemiDerzsy Website: http://www.noemiderzsy.com