How to Scrape Tweets Data by Location Using Python and snscrape?

Introduction of snscrape Using Python

In this blog, we’ll use tahe development version of snscrape that can be installed with

pip install git+https://github.com/JustAnotherArchivist/snscrape.git

Note: this needs Python 3.8 or latest

Some familiarity of the Pandas module is needed.

Three packages are available given below

To get the primary (i.e. most current) 100 tweets which contains phrase data science, we can utilize the code:

That can be shortened into given line:

Outputting the initial five results, we can start seeing the information that line provides us:

however this isn’t it! It gives 21 data columns in reply including:

We encourage you to explore and experiment with the various features of snscrape to better understand its capabilities. Additionally, you can refer to the mentioned article for more in-depth information on the subject. Later in this blog, we will delve deeper into the user field and its significance in tweet scraping. By gaining a deeper understanding of these concepts, you can harness the full potential of snscrape for your scraping tasks.

Advanced Search Features

When it comes to scraping tweets by location using snscrape, you have two options: utilizing the "near:city" tag along with "within:radius" or using "geocode:lat,long,radius." Through thorough research, it has been confirmed that these options yield identical results when used correctly, as interpreted by Twitter.

In this code snippet, we define the search query as "pizza near:Los Angeles within:10km", which specifies that we want to search for tweets containing the word "pizza" near Los Angeles within a radius of 10 km. The TwitterSearchScraper object is created with the search query, and then we iterate over the retrieved tweets and print their content.

Feel free to adjust the search query and radius per your specific requirements.

For comparing results, we can utilize an inner merging on two DataFrames:

common_rows = df_coord.merge(df_city, how='inner')

That returns 50 , for example, they both have the same rows.

What precisely is this place or location?

When determining the location of tweets on Twitter, there are two primary sources: the geo-tag associated with a specific tweet and the user's location mentioned in their profile. However, it's important to note that only a small percentage of tweets (approximately 1-2%) are geo-tagged, making it an unreliable metric for location-based searches. On the other hand, many users include a location in their profile, but it's worth noting that these locations can be arbitrary and inaccurate. Some users provide helpful information like "London, England," while others might use humorous or irrelevant descriptions like "My Parents' Basement."

Despite the limited availability and potential inaccuracies of geo-tagged tweets and user profile locations, Twitter employs algorithms as part of its advanced search functionality to interpret a user's location based on their profile. This means that when you look for tweets through coordinates or city names, the search results will include tweets geotagged from the location and tweets posted by users who have that location (or a location nearby) mentioned in their profile.

Twitter's advanced search algorithms consider geo-tagged tweets and user profile locations to provide a broader set of tweets when performing location-based searches.

To illustrate the usage of location-based searching on Twitter, let's consider an example. Suppose we perform a search for tweets near "London." Here are two examples of tweets that were found using different methods:

The first tweet is geo-tagged, which means it contains specific geographic coordinates indicating its location. In this case, the tweet was found because of its geo-tag, regardless of whether the user has a location mentioned in their profile or not.

The following tweet isn’t geo-tagged, which means that it doesn't have explicit geographic coordinates associated with it. However, it was still included in the search results because a user has given a location in the profile that matches or is closely associated with London.

When performing a location-based search on Twitter, you can come across tweets that are either geo-tagged or have users with matching or relevant locations mentioned in their profiles. This allows for a more comprehensive search, capturing tweets from specific geographic locations and users who have declared their association with those locations.

Get Location From Scraped Tweets

If you're using snscrape to scrape tweets and want to extract the user's location from the scraped data, you can do so by following these steps. In the example below, we scrape 50 tweets within a 10km radius of Los Angeles, store the data in a DataFrame, and then create a new column to capture the user's location.

You can customize the code further to suit your needs, such as extracting additional tweet data or analyzing the scraped tweets and user locations. By iterating over the scraped tweets, you can access the user.location attribute to retrieve the user's location information. This value is then stored in a new column called "user_location" in the DataFrame.

Upon inspecting the first 5 rows of the DataFrame, it is evident that while the location formats vary, they can all be interpreted as referring to Los Angeles.

If It Doesn’t Work According to Your Expectations

The use of the near: and geocode: tags in Twitter's advanced search can sometimes yield inconsistent results, especially when searching for specific towns, villages, or countries. For instance, while searching for tweets nearby Lewisham, the results may show tweets from a completely different location, such as Hobart, Australia, which is over 17,000 km away.

To ensure more accurate results when scraping tweets by locations using snscrape, it is recommended to use the geocode tag having longitude & latitude coordinates, along with a specified radius, to narrow down the search area. This approach will provide more reliable and precise results based on the available data and features.

Conclusion

In conclusion, the snscrape Python module is a valuable tool for conducting specific and powerful searches on Twitter. Twitter has made significant efforts to convert user input locations into real places, enabling easy searching by name or coordinates. By leveraging its capabilities, users can extract relevant information from tweets based on various criteria.

For research, analysis, or other purposes, snscrape empowers users to extract valuable insights from Twitter data. Tweets serve as a valuable source of information. When combined with the capabilities of snscrape, even individuals with limited experience in Data Science or subject knowledge can undertake exciting projects.

Happy scrapping!

For more details, you can contact Actowiz Solutions anytime! Call us for all your mobile app scraping and web scraping services requirements.