Code used for reading, counting, and pseudo-measuring the sentiment of environmentally sustainable words in company 10-K annual statements from the SEC. The program mainly uses the NLTK to extract words of interest (concept vectors) and also sentences that contain these words for 30 Fortune 500 companies.
getConceptVectors: The program uses concept vectors or words that are of interest to search through the annual statements (10-K). This function extracts the Concept Vectors from a text file stored on the local computer.
calculateWordFreq: This function calculates the absolute frequency of the concept vectors in each of the 10-Ks that are being examined.
calculateNormWordFreq: This function calculates the normalized frequency of the concept vectors in each of the 10-Ks that are being examined.
writeCSVWordFreqScores: Writes the frequncies into a .csv file.
sentExtract: This function extracts the sentences within which the concept vectors occur.
getScores: This function imports the 2010 Newsweek sustainability scores from a .txt file.
matchScores: The scores are matched to a given company (one of 30) that are being examined
writeCSVRegression: The function writes a .csv file that is later used to regress the frequency of sustainable words against the Newsweek sustainability score.
sentenceSentiment: A pseudo sentiment analysis function that counts the frequency of positive and negative words in the sentences that contain the concept vectors (positive and negative word lists downloaded from: http://nealcaren.web.unc.edu/an-introduction-to-text-analysis-with-python-part-3/)
findIndustry: Extracts the industry of each of the companies from the Newsweek sustainability scores .txt file so that use of sustainable words can be generalized to the industry level.
writeCSVSentenceSentiment: A .csv file is created with the company name, industry, psotive, and negative word frequencies for the sentences extracted.