1.

Introduction
For data analytics is one of the tools business management now comes for analysing trends,
optimization of operations, and customer experience. Among the leading platforms for short-
term rentals, Airbnb generates huge amounts of datasets containing unprecedented amount of
information for studying pricing, occupancy and behaviour of customers. But it’s only after this
step that meaningful analysis is possible, before that, the dataset must be clean, structured and
also without inconsistencies.

The incorrect or missing data can generate misleading insights, which subsequently will
negatively affect the business strategy. As such, data preprocessing is a necessary step to assure
high quality data. The aim of this report is on how to analyses the Barwon Southwest, Victoria,
Australia dataset provided by Inside Airbnb. According to the major objectives of the report,
they are to:

1. Identify common data quality issues.

2. Try to find appropriate patterns or rules to strengthen data reliability.

3. Perform data preprocessing steps to be able to work with the dataset later.

Microsoft Excel will process the dataset, handling the missing value, remove duplicates, remove
inconsistency, and find outlier. Quality of data will ensure insights that can drive meaningful
business decision in the short term rental market.

2. Working with Data Quality Issues in Airbnb Dataset

Quality of any dataset is required to hold and being complete, and accurate and consistent. A key
data quality issue found in the Airbnb dataset that if not addressed can lead to error of analysis
and faulty decision making is identified.

2.1 Missing Values

Issue
Large datasets often face the problem of missing values especially when users need to provide
input information. The crucial fields of our dataset contained significant missing data points
among them:

 Last_review does not show any recorded dates for numerous properties within the
dataset. Data collection errors or properties without reviews seem to be the cause of these
missing values.
 The reviews_per_month column shows several empty values since some properties do
not receive regular reviews.
 Several properties have missing price information which negatively affects revenue
measurement computations.

Impact on Analysis
Time-based tracking of customer engagement becomes challenging because last_review values
are missing from some entries.

 No reviews_per_month values means a listing’s popularity cannot be measured properly.

 The absence of prices results in erroneous calculations of revenue which hinders general
market analyses and rental pricing strategy evaluation.

2.2 Duplicate Entries

Issue

 The occurrence of duplicate records is a common problem which researchers detect

during database analysis especially when these records get created due to:
 Multiple listings of the same property by hosts occur across various differently named
listings.
 Repeated rows appear in the database because of data retrieval problems.

Impact on Analysis

 When listing counts are inflated it misleads stakeholders about the real number of
available units since the statistics become misleading.
 The combination of double-counted revenue projections occurs because of duplicate
records that create incorrect revenue estimates.
  Statistical model biases produce wrong price recommendations which make negative
impacts on business choices.

2.3 Inconsistent Data Formats

Issue

The analysis becomes difficult due to inconsistent data formats since they prevent effective
numerical operations and date calculations. The following problems can be found in our dataset:

 Price data exists as text in the values "$120" without proper numeric representation (120).
 The data dates appear in two different formats ("2023-12-01" and "Dec 1, 2023")
preventing effective date-based analysis or sorting.

Impact on Analysis

 The non-standardization of prices makes mathematical computations impossible such as

determining nightly average rates.
 The mixture of different date formatting methods prevents suitable analysis of rental
pattern forecasts across time periods.

2.4 Outliers and Anomalous Data

Issue

 The data contains values or observations which stand distinctively separate from the
normal range of expected values. Two common outlier types present in the data set
involve:
 Fictitious rental options exist because some listings set minimum_nights values at
numbers greater than 365 days.
 Data containing impractical nightly rates ($10,000) at extreme values probably result
from user entry mistakes.

Impact on Analysis

 Unusual points affect statistical averages and group patterns so business analytics might
show incorrect information.
  When pricing data deviates from reasonable standards it causes problems in financial
forecasting and competitive market assessments to go wrong.

3. Data Cleaning Techniques

Several data cleaning techniques were applied to improve the dataset’s reliability and accuracy
because of identified data quality concerns.

3.1 Handling Missing Values

The analysts used appropriate handling techniques which depended on data types to deal with
missing values because these values could create analysis problems.

Techniques Used:

 When dealing with numerical data fields like reviews_per_month the missing values
received replacement using the median value from the specific column. Data points
surrounding the median act as protection against extreme values that would distort the
dataset.
 The category field last_review received "No Review" as a value when missing data was
found to indicate the listing lacked customer reviews.

Application in Airbnb Dataset:

All empty reviews_per_month entries received new values from the median measurement of
available review information. To achieve data consistency all missing last_review fields received
a value of "No Review."

3.2 Removing Duplicates

Recognizing duplicate data remains essential because retaining such data produces incorrect
analytics so the user needs to remove them.

Techniques Used:

Excel Remove Duplicates identified and eliminated duplicate records from the table.

Application in Airbnb Dataset:

  A duplicate check was applied to the id column for determining repeated listings.
 A process was conducted to eliminate listing entries with matching name along with
identical host_id and price data points thus maintaining one unique entry for each listing.

3.3 Handling Inconsistent Data Formats

All values received standardized treatment as a means to ensure data consistency.

Techniques Used:

 The price text was converted into numeric format through a process of dollar sign ($)
deletion.
 We applied Excel’s Power Query Editor to convert all last_review data into the YYYY-
MM-DD date format with standardized consistency.

Application in Airbnb Dataset:

The Find and Replace functionality in Excel allowed for removing currency symbols ($) from
the price data field. The dates were reformatted for unified presentation within the entire record
set.

3.4 Identifying and Removing Outliers

We employed filtering methods to find extreme values because they tend to contaminate analysis
then we removed them.

Techniques Used:

Z-score analysis was applied for detecting extreme outlier points. The process excluded improper
entries that showed minimum_nights values greater than 365 days.

Application in Airbnb Dataset:

 The data science analysis excluded bookings with minimum_nights set to more than 365
days since these conditions indicate unfeasible availability.
 Properties with nightly pricing above $3000 were filtered from the dataset to eliminate
wrong values.
4. Data Preprocessing Steps
The manipulated data needed preprocessing to achieve suitable structure for examination after
completing raw data cleaning. The administration of data precedes analysis since it reforms data
standards and builds precision while removing inconsistencies that affect the clarity of insights.
The preprocessing process describes the main steps used on the Airbnb dataset which involved
selecting essential columns and missing value handling and data format conversion before
creating a saved database for analytical use.

4.1 Selecting Relevant Column

The Airbnb dataset included multiple columns but most of them were not vital for research
purposes. The dataset required simplification so only fundamental columns related to analysis
were kept for further assessment. The chosen columns went through evaluation to determine
their value in pricing, room type, availability and user engagement measurements. The chosen
set of columns included these characteristics:

 Each listing contains a title named in the Airbnb database which serves to describe its
accommodation type.
 A rental space determination decides either a guest experiences an entire home or rents a
private space or shares accommodations with others.
 The nightly rental fee represents the price that customers need to pay for booking the
accommodation.
 Guests need to reserve their stays for a minimum number of continuous nights according
to this attribute.
 Reviews per month indicates the standard number of monthly reviews which a listing
generates.
 The available booking duration throughout a year equates to availability_365.

Rationale for Column Selection

The evaluation of pricing trends together with rental type distribution requires this information.
The availability feature enables an understanding of how frequently properties become occupied
and thus facilitates revenue projection.
A record of the minimum required nights in a stay enables the evaluation of rental rules along
with market vitality.

Reviews Per Month: Serves as an indicator of a listing’s popularity and customer satisfaction.

The removal of host_name and scrape_id metadata columns alongside other irrelevant data
columns optimized the dataset for effective analysis and visualization.

4.2 Cleaning the Dataset

The dataset required cleaning operations which included the removal of unneeded data while
fixing inconsistent values along with correct formatting throughout all records. The primary
cleaning steps included:

 The analysis excluded unnecessary features host_name and scrape_id from the dataset.
The examined dataset became harder to analyze and more complex through these
unneeded columns that provided no core analysis value.
 The cleaning procedure standardized the naming conventions of the room_type category.
Different spellings of "Entire home" such as "Entire Home" and "entire home" were
converted into a unified term.
 Some listings contained insupportable minimum_nights values above 365 days so these
records were deactivated from the dataset.
 The cleaning process of the dataset became essential for achieving precise outcomes in
subsequent price predictions together with demand forecasting and trend analysis.

4.3 Handling Missing Data in Detail

The correct management of missing data proves to be essential during preprocessing because it
prevents unreliable models and wrong conclusions from forming. The approach to handle
missing values utilized separate techniques depending on whether the data consisted of
numerical or categorical fields.

Approach for Numerical Data

Interpolation methods were used to perform estimations for the missing review_per_month data
points. Data trends can be preserved through this approach which helps prevent sudden changes
in the data distribution pattern. The median value from numerical data points was substituted for
missing values in reviews_per_month to prevent wrong outcome results caused by outlying data
points.

Approach for Categorical Data

The field containing missing values in last_review received "No Review" as default value to
indicate listings which had not been reviewed.

Standardization of unknown data involved assigning "Unknown" as the default value for
room_type fields so that data maintenance could continue without information loss.

Systematic handling of missing values kept all important dataset information intact while
maintaining validation for analysis purposes.

4.4 Transforming Data Formats

The process of data format transformation establishes proper value structures needed for
analysis. Standardization routines were applied to numerical and categorical data fields of the
Airbnb dataset.

Standardizing the Price Column

The price column had its initial data format as text which contained dollar signs ($138)
throughout its numerical values. To convert the column into numbers only we used the Find and
Replace function in Excel for dollar sign elimination. Performing this step enabled the system to
perform mathematical computations for average pricing analysis and price pattern examination.

Standardizing Date Formats

The date information in the last_review column showed multiple data formats between "2023-
12-01" and "Dec 1, 2023". Power Query in Excel applied the YYYY-MM-DD format to
restructure all date values. The standardization approach made it possible to execute date-related
analyses including time-series forecasting and review trend analysis without facing any
inconsistencies. The process involves transforming all disparate categorical data labels into
standardized labels. The room_type column included multiple variants of equivalent category
types which included "Entire home" and "Entire house". Standardization processes unified all
labels so duplicate categories would not arise during analysis. The dataset required data
transformation before analysis because it needed proper formatting to succeed with computations
and visualization need.

4.5 Exporting and Storing the Processed Dataset

The cleaned pre-processed data received organized storage through a specified format that would
support upcoming analysis needs.

Steps Taken for Storage:

 A CSV file was chosen for the cleaned dataset's storage because it enables compatibility
with Excel applications.
 A complete set of backup versions maintained the original dataset for future verification
and comparison operations.
 The documentation process introduced new documentation that explained every
preprocessing step together with modification descriptions and data transformation
reasons.
 Data storage of processed data in a well-structured format made it possible to perform
additional analysis and verify research results.

5. Significance of Data Cleaning and Preprocessing

The process of data cleaning together with preprocessing transforms data quality into improved
operational performance and produces better analytical results. The below list demonstrates the
essential advantages for data cleaning and preprocessing the Airbnb dataset:

5.1 Ensuring Accurate Insights

Data cleaning generates reliable information that delivers accurate insights which businesses can
use for making decisions based on trustworthy data.

Data integrity improves because these processes eliminate duplicate information while fixing
data mistakes, so the analysed results match actual industry patterns.

5.2 Enhancing Efficiency in Data Analysis

Preprocesses data streamlines data maintenance tasks because analysts can dedicate their time to
extraction rather than error correction. A properly organized dataset makes computations run
more efficiently so both machine learning algorithms and statistical analysis take reduced
execution times.

5.3 Improving Machine Learning Model Performance

Better predictions together with more accurate models result from high-quality data processing.
The process of handling missing values along with removing outliers ensures machine learning
algorithms will not learn from unjustified and faulty data points. The preprocessing of data leads
to better analytical results and establishes fact-based and accurate data-driven decisions.

6. Conclusion
The examination stressed how important data preparation methods are to prepare a dataset for
analytical work. The key data quality problems we found existed in four domains such as
missing values among other issues including duplicate entries accompanied by inconsistent
formats along with outliers. We performed different data cleaning procedures to resolve the
listed data challenges by removing duplicated records while dealing with missing values and
converting formats to standard types and removing extreme outlier points.

The processing phase included the following steps:

 Selected relevant columns to streamline analysis.

 The data required transformation of numerical variables and categorical values to achieve
uniformity within the dataset.
 The processed dataset was exported for visualization analysis.

The cleansed dataset enables users to both predict business trends and build models as well as
create appealing visualizations. The team should task itself with implementing machine learning
methods to recognize property pricing trends while developing optimal rental pricing models.
The quality of input data forms the essential basis for acquiring meaningful insights from
datasets because it enables effective extraction of practical knowledge from original data
sources.