Monthly US Food Recall Data with US state and food recall categories

Access to the food recall data from the FDA can be helpful for anyone looking to stay up-to-date on food safety issues or for businesses in the food industry looking to ensure compliance with regulations. The data includes information on recalled products and the reasons for the recall. The author has made sure that the area of food distribution is represented in a list of states (e.g. ['TX', 'CA', 'WA', 'IL', 'FL', 'PA', 'NY', 'MD', 'CO'] ). Also the recalls have been linked to 6 categories (e.g. Allergens, Bacteria, Temperature, Toxins, Foreign Objects and Mislabeled) using a ‘bag-of-words’ technique. The data is now in better shape for analysis by location and type of recall. All original data is included in the data set. (written by AI in Notion: edited by Alicia Powers)

Additional Information About Food Recall Enforcement Reports

“When an FDA-regulated product is either defective or potentially harmful, recalling that product—removing it from the market or correcting the problem—is the most effective means for protecting the public.

Recalls are almost always voluntary, meaning a company discovers a problem and recalls a product on its own. Other times a company recalls a product after FDA raises concerns. Only in rare cases will FDA request or order a recall. But in every case, FDA's role is to oversee a company's strategy, classify the recalled products according to the level of hazard involved, and assess the adequacy of the recall. Recall information is posted in the Enforcement Reports once the products are classified.”

(from FDA website: https://open.fda.gov/apis/food/enforcement/ )

The original data can be found here:

There are 1 files, last updated on 2023-03-02.

• /food/enforcement data4.58 mb

https://open.fda.gov/data/downloads/

What you get:

data_of_data_download="March 2, 2023"

*******This code is a brute force method to categorize reasons for recall and clean up the state distribution data. Bag of words approach used to create categories. These categories have not been finalized with an expert in food alerts.

*******The code could be 'fancier' using natural language analysis (NLTK) and ML model to predict categories based on bigrams and text snippets in data.

********Example food distribution location data

Problem: states not standardized or in a list

ORIGINAL DATA examples for food distribution location:

Distributed in CA, CO, FL, IL, MD, NY, PA, TX, and WA.

Distributed in Montana and Washington.

Output after code: ['TX', 'CA', 'WA', 'IL', 'FL', 'PA', 'NY', 'MD', 'CO']

*********Example recall reasons data are categorized in the following categories: Allergens, Bacteria, Temperature, Toxins, Foreign Objects, and Mislabeled

Allergens: focuses on food allergies and bodily reactions to food

(variables: allergies_list, allergies_bin)

Bacteria (not verified by a scientist): focuses words associated with bacteria

(variables: bacteria_list, bacteria_bin)

Temperature: focused on pasteurization, process and temperature related recalls

(variables: temp_list, temp_bin)

Toxins (not verified by a scientist): focused toxins like pesticides

(variables: toxins_list, toxins_bin)

Foreign Object: focused on foreign objects like metal and plastics

(variables: foreign_object_list, foreign_object_bin)

Mislabeled: focused on mislabeled recalls

(variables: mislabeled_list, mislabeled_bin)

Problem: hard to digest recall information without reading every description

ORIGINAL DATA examples:

Undeclared Peanut.

Plastic fragments. There is potential for Lexan Plastic to be in the product.

Raw material tested positive for Cronobacter sakazakii.

Undeclared FD&C Yellow No. 5

Additional columns are added to list possible allergins, toxins, bacterias, etc. and a binary columns that says whether this recall is related to possible allergins, toxins, bacterias, etc.

Original recall data: Undeclared Peanut.

Output after code (variable: allergies_list): ['peanut', 'nut', 'undeclared']

Output after code (variable: allergies_bin): 1