Problem Statement

• HealthCare and medical centers are going through crisis.
• This analysis will majorly focus on what are the core reasons for less ratings and prediction of highest and lowest levels of parameters contributing in growth of a business.
• Summarizing how a business in crisis should take the advantage of the flaws of flourishing businesses.

DATASET UNDERSTANDING

Business dataset – Contain details of all the businesses

Review dataset – Reviews given by users

User dataset – Contains user’s and metadata associated with users

Check-in data – Data of business check-ins

Tip data - Tips written by a user on a business

​

Medical Dataset – Contains all the details businesses and reviews in HealthCare and Medical category.

173941 12

5261668 9

3911218 4

1326100 22

1098324 5

ROW COLUMN

49041 20

Analysing Dataset

RELAVENT ATTRIBUTES FOR HEALTH-CARE AND MEDICAL CENTRE

• Business_id - For identifying different businesses
• City – For analysis of particular city
• Latitude & longitude- For location of business
• Stars – Ratings given by users
• Review_count - Total reviews received by business
• Category – For selecting type if business
• Weekday – For analyzing days with maximum output
• Text - Review given by user
• Hours- Timing of visits in a business

​

​

​

YELP DATASET ALL BUSINESS CATEGORY

IN-DEPTH INVESTIGATION OF HEALTH-CARE AND MEDICAL BUSINESSES

TOP CATEGORIES IN HEALTHCARE AND MEDICAL DATASET�

​

• DOCTORS
• DENTISTS
• BEAUTY AND SPA
• COSMETIC DESTISTS
• MASSAGE THERAPY
• MEDICAL CENTRE

DATA-PREPROCESSING

MAP OF NULL VALUES

INFERENCE REGARDING NULL VALUES

• Neighbourhood attribute column was dropped from the dataset.
• Rows where dropped of other null values.

​

Attribute with more than 90% of null values

Neighbourhood

Attribute with negligible amount of null values

Postal Code

REMOVING OUTIERS�

BEFORE

V/S

AFTER

INFERENCE REGARDING REMOVING OUTLIERS

• From the previous graph we can see before and after removing null values
• Business dataset containing review count attribute had most number of outliers which can be removed

ANALYSIS AND IMPORTANT QUESTIONS IDENTIFIED

PARKING DIFFERENCE BETWEEN A LESS AND A HIGH RATED MEDICAL CENTERS

LESS RATED MEDICAL CENTRE

HIGH RATED MEDICAL CENTRE

High rated medical centres

• Many parking areas around the medical center
• They can park anywhere apart from the medical center itself

​

Less rated medical centres

• Small and only single parking space in the area
• Loss of potential customers

ANALYSIS OF PARKING DIFFERENCES

INFERENCE FROM THE MEDICAL CATEGORY

Patient visits medical centre

Did they gave review?

Is it good or bad?

Changes needed

Loss of potential good ratings

Rating increases

Good

Bad

AREA OF FOCUS

​

Might not write a review if everything goes well maybe because they are unwell

​

​

WORDS USED BY REVIEWERS FOR LESS RATED AND HIGH RATED MEDICAL CENTERS

LESS RATED

HIGH RATED

Great: We saw a great working culture

Dentist : The dentist was highly skilled

Experience: The doctors have lot of experience

Waiting: We were waiting in the lobby for long time

Aggressive: The employees are very aggressive

Appointment: The doctors have lot of experience

HOW THE LOCATION IS CAUSING A CRISIS IN A HEALTHCARE CENTRE?

• Less rated are in outskirts of city
• In urban places surrounded by high rated centers
• Mostly ignored in infrastructural functioning.
• Remote centers are not rated well

​

Initial Analysis

STAR MAP OF HEALTHCARE CENTERS

STAR MAP OF HEALTHCARE CENTERS

LESS RATED MEDICAL CENTERS

ISSUE

​

1. Medical centres at outskirts of city with less rating

​

• Medical centres at urban places with less ratings

PURPOSED SOLUTION

• Pathway and quick accessibility for potential customers

​

• Working on flaws of high rated centres

​

INFERENCES FROM THE STAR MAP

HEALTHCARE CENTRE CHECKIN TIMING ANALYSIS

Frequent visits at 10 AM

• Early morning work
• Can take leave for entire day

​

Less visits on Thursday

• Middle of the week
• Slow day with less patients

INFERENCES AFTER ANALYSING CHECKIN TIMINGS

WHAT ARE THE SHORTCOMINGS OF GOOD RATED MEDICAL CENTRE WHICH WILL BE AN ADVANTAGE FOR THE BUSINESS IN CRISIS?�

Flourishing businesses

Business in Crisis

Fremont children clinic

They have been on time or even faster than that. I will reduce a star though because there is never any parking.

Sam’s skin care

I love this practice, and they're one of the few places that does simple blood work, but I'm not sure about the hygiene at this clinic.

Pueblo Medical Imaging

Ed De Andrade, DDS

 An excellent chairside manner, but the billing office has been the worst experience

​

PARKING

​

​

​

HYGINE

​

​

​

BILLING

�PREDICTING THE FACTORS WHICH WILL HELP IN IMPROVING THE HEALTHCARE CRISIS USING NATURAL LANGUAGE PROCESSING (NLP)�

PREDICTING MODEL USED : MULTINOMIAL NAIVE BAYES

​

• Prediction of highest and the lowest levels of ratings using the review.
• Frequency of words impacting the medical centre.

​

​

Word frequency of Less ratings

Word frequency of high ratings

Factors on which healthcare centres need to work on their business

​

​

Keeping factors which cause high ratings

Improving on factors which are causing less ratings

EXTRACTED WORDS USING NLTK

• APPOINTMENTS
• DOCTORS
• STAFF
• INSURANCE

• DENTIST
• THERAPIST
• EMPLOYEE SKILLS
• BILLING

PRECISION RESULTS

• Prediction of highest and lowest levels with very high accuracy
• Keeping things when it gets high ratings
• And improving when it gets less ratings

​

​

CONCLUSION

• For Medical centers of Yelp, appointments, location and doctors mean everything.
• No matter what star one kinds of medical centers get, more frequent words are appointments, doctors and some other words (hygiene, staff etc.)
• Therefore, all medical centers should other factors and just focus on improving appointments and having qualified doctors.

OUTCOMES

​

• From the analysis of this dataset when got to know about the reasons for the crisis in a health case business.
• We also got to invest time in our technical and non technical skills.

​

​

FUTURE SCOPE

Working with MLOPS (Machine learning operations)

Continuous integrations

Continuous Delivery

Continuous Deployment

Deployed in cloud accessible to anyone

References

• The Yelp dataset is a subset of our businesses, reviews, and user data for use in personal, educational, and academic purposes. https://www.yelp.com/dataset

​

• NLTK is a leading platform for building Python programs to work with human language data. It provides easy-to-use interfaces to over 50 corpora and lexical resources such as WordNet, along with a suite of text processing libraries for classification, tokenization, stemming, tagging, parsing, and semantic reasoning, wrappers for industrial-strength NLP libraries

​

• Multinomial Naive Bayes https://scikit-learn.org/stable/modules/generated/sklearn.naive_bayes.MultinomialNB.html

​

• Yelp Doctor Reviews: The Pitfalls to Avoid When Choosing a Care Provider https://www.medifind.com/news/post/yelp-doctor-reviews

​

THANK YOU