Data exploration is not merely a preliminary step in the analysis process; it is a fundamental pillar that lays the groundwork for uncovering hidden insights, revealing intricate patterns, and extracting valuable knowledge from raw datasets. At its core, data exploration is about delving into the depths of data, unraveling its mysteries, and unlocking its potential. It empowers us to ask the right questions, uncover unexpected correlations, and gain a deeper understanding of the underlying phenomena captured within the data.

Through meticulous examination and analysis, data exploration sheds light on the intricate interplay between variables, uncovering hidden patterns, trends, and outliers that may have gone unnoticed. It fosters a deeper appreciation for the richness and complexity of data, revealing its nuances and intricacies.

Let's begin our exploration into the world of data analysis and uncover the journey we undertook to accomplish this task.

Dataset before cleaning:

The dataset obtained from the website after web scraping exhibits various inconsistencies and inaccuracies, as highlighted in the figure provided. One noticeable issue is the presence of an extraneous column labeled "Unnamed: 0," which merely contains indices for each record in the dataset and does not contribute any meaningful information. Furthermore, columns such as "Name," "developers_publishers," and "genre" contain entries formatted as strings with unnecessary elements like square brackets and single quotes, which can impede data analysis.

Additionally, the "release_date" column consists of entries in string format, hindering the dataset's usability and analytical capabilities. Converting these entries into datetime format would enhance the dataset's reliability and facilitate more comprehensive data analysis and interpretation. Similarly, entries in the "price" column are currently stored as strings, and converting them to a float data type would improve computational efficiency.

One notable issue is the substantial number of missing values present in the "playtime_total" column, totaling over 8000 entries. This prevalence of missing data can significantly impact the analysis of the dataset, potentially leading to erroneous interpretations. Therefore, it is crucial to address these missing values meticulously to ensure the accuracy and reliability of the dataset for subsequent analysis.

Dataset after cleaning:

Let's break down each step and explain how the dataset is cleaned:

• Dropping Columns with Null Values:

The first step involves dropping the column "playtime_total" since it contains a lot of null values. This is done using the `drop()` function with the `axis=1` parameter to specify column-wise dropping.

• Dropping Rows with Null Values and Adjusting Indexes:

 Next, rows with null values in any column are dropped, and the indexes are reset to ensure continuous indexing using the `dropna()` function. The `reset_index(drop=True)` part resets the index while dropping the old index.

• Dropping Unnamed Column:

The column labeled "Unnamed: 0" is dropped using the `drop()` function with the `columns` parameter specifying the column name and `axis=1` indicating column-wise dropping.

• Cleaning String Entries:

Entries in the "Name," "developers_publishers," and "genre" columns undergo string cleaning to remove square brackets, double quotes, and single quotes. This is achieved using the `str.replace()` function to replace specific characters with an empty string or to manipulate strings using NumPy's `np.where()` function.

• Splitting and Rearranging Developers and Publishers:

Entries in the "developers_publishers" column are split into separate lists for developers and publishers using a loop. The split text is then assigned to separate lists (`dev_list` and `pub_list`) and subsequently added as new columns ("Developers" and "Publishers") to the dataset. The original "developers_publishers" column is dropped using the `drop()` function.

• Converting Date Entries to Datetime Format:

Entries in the "Release_Date" column are converted from string format to datetime format using the `pd.to_datetime()` function. The `format` parameter specifies the format of the date string, and `errors='coerce'` is used to handle any conversion errors gracefully by setting invalid parsing to NaT (Not a Time).

• Converting Price Entries to Float Data Type:

Entries in the "Price (in $)" column are converted from string format to float data type by removing the dollar sign ('$') using string manipulation and then converting the values to float using `astype(float)`.

• Handling Owners Column:

Entries in the "Owners" column are cleaned by removing extraneous characters like '\xa0..\xa0' and replacing them with a hyphen ('-'). This is achieved using list comprehension and string manipulation.

Overall, these cleaning steps address various inconsistencies and inaccuracies in the dataset, ensuring its usability, reliability, and suitability for further analysis. Each step targets specific data quality issues and employs appropriate techniques to clean and transform the dataset effectively. Below is a snapshot of the cleaned dataset.

Dataset after cleaning:

Data Visualization

By these Visualizations, We can assume what is the trend of the well selling games. 

• From Image 1, We can see the pie of the games about prices. 

• From Image 2, There are almost 30% of games are Indie. 

• From Image 4, We can find what is most popular game. 

• From Image 5, September is the biggest month that games are released.

• From Image 6, Most of developers/publishers released one game per year.

• From image 8, Still the Famous developer/publisher has most followers.