File size: 7,165 Bytes
d8125a9 7643fb4 327ce35 7643fb4 81b3c1d 7643fb4 f246bf7 7643fb4 4a920ff 7643fb4 4a920ff 7643fb4 0501d64 7643fb4 35d30e1 7643fb4 4a920ff 7643fb4 b5bca16 35d30e1 e47b774 cc694c2 bd4110c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 |
# EDA Summary — Airbnb Open Data - Lior Feinstein
This project explores an Airbnb dataset to understand which factors influence the review ratings that listings receive. The analysis includes data cleaning, visualizations, and investigations into how price, room type, neighbourhood group, and host activity relate to guest review rating.
## 1. Dataset Overview
The dataset consists of 102,599 Airbnb listings in the New York City,United States. It includes 26 original features covering listing attributes, host characteristics, availability details, and review metrics.
### Key Feature Types (full list in the notebook)
**Categorical Features:**
- `room_type`
- `neighbourhood_group`
- `host_identity_verified`
- `cancellation_policy`
- `instant_bookable`
**Numerical Features:**
- `price`
- `service_fee`
- `review_rate_number`
- `minimum_nights`
- `number_of_reviews`
- `availability_365`
### Prediction Goal
The primary task is a regression problem:
Predict the numeric `review_rate_number` (1.0–5.0) for each listing based on features such as price, room type, and neighbourhood group.
---
## 2. Data Cleaning
A series of preprocessing steps were performed to prepare the dataset for analysis and modeling.
### Columns Removed
- `host_id`, `host_name`, `name` — Identifier/text fields not useful for prediction
- `country_code`, `country` — All entries are within the US
- `cancellation_policy`, `house_rules` — Non-numeric and removable per instructions
- `last_review` — Not relevant to the analysis
- `license` — Mostly empty
### Data Quality Fixes
- Corrected spelling issues in `neighbourhood_group`:
- "brooklm" to "Brooklyn"
- "manhatan" to "Manhattan"
- Handled rare categories by checking frequency
- Filled missing values
- Resolved contradictions (e.g., listings with 0 reviews but a rating)
- Removed duplicate listings using the `id` field
### Outliers handling
I handled outliers using box plots:
it is noticable that minimum nights has outliers as it is not possible to have negative value, and it is weird to have values larger than 365.
**number of reviews and reviews per month** : we can see there are a lot of outliers.
That popular listing is an outlier, but it's one of the most important data points! We definitely keep these, it just implies about the property popularity.
**calculated host listings count** - there are many "Far" data point but, it might be "superhosts" or property managers who own many listings. we will keep it.
---
## 3. Visualizations and Insights
The EDA included various visual explorations of the dataset to examine relationships between listing characteristics and review ratings.
The prices per room type varies, and there doens't seem to be an outlier. no changes will be done to the data.
---
## 4. Research Questions and Findings
### Q1 — Is Room Type Related to Being Highly Rated?
Room type shows only a weak relationship with high ratings.
- Hotel rooms have the highest proportion of 5-star reviews
- Shared rooms have the lowest
- Overall, differences across room types are modest, suggesting room type alone is not a strong predictor of guest satisfaction
---
### Q2 — What Is the Price Distribution of Highly-Rated vs. Not-Highly-Rated Listings?
Key observations:
- Listings with a rating of 0 tend to have much lower median prices
- This is likely because new or unrated properties set lower prices to attract initial guests
- Once listings begin receiving ratings, their prices increase and become more consistent
This suggests that hosts raise prices after building credibility through reviews.
---
### Q3 — Does Neighbourhood Group Affect Rating?
There is an important difference between raw counts and percentages:
- Manhattan has the highest number of 5-star reviews because it has the most listings
- Staten Island and Queens have the highest proportion of 5-star ratings
This indicates that while Manhattan has high volume, its rating consistency is lower. In contrast, smaller markets like Staten Island show more uniform guest satisfaction.
---
### Q4 — Does the Number of Listings a Host Manages Influence Their Ratings?
A clear but slight pattern emerges:
- Hosts who manage more properties tend to have higher average ratings
This may suggest that professional or experienced hosts provide more consistent service and maintain higher-quality listings.
## Final Summary:
- Room type has only a minor impact on ratings, with small differences across room categories.
- Price and rating show no clear correlation, and low prices mainly indicate new, unrated listings.
- Neighbourhood group influences rating consistency, with Staten Island and Queens performing best proportionally.
- More experienced hosts (managing more listings) consistently receive higher ratings.
## Recommendations:
- To better understand rating patterns, Airbnb should collect additional guest-experience metrics such as cleanliness, communication, response time, and review sentiment, since these factors likely drive satisfaction more than listing attributes.
- Because experienced hosts receive higher ratings, Airbnb should also offer training and support programs—such as onboarding guides, communication templates, and best-practice recommendations—to help new hosts deliver more consistent, high-quality service.
Video Presentaion Link:
https://drive.google.com/file/d/1INd5mj2PZmQ2AeaZzTGegr71bwfOGU2u/view?usp=sharing |