Charts from Twitter Data — A Practical Guide
Twitter (X) data is some of the most chartable social media data — high volume, timestamped, with engagement signals on every post. Common charts people build: tweet volume by hour, sentiment over time, follower growth curves, hashtag adoption rates, retweet network graphs, top-account engagement comparison.
Three paths to build these: free tools like Vidooly's free dashboards, public Kaggle datasets (great for one-shot research, stale for real-time), or your own pipeline calling an API and pushing data into a charting layer. The first two work for retrospective analysis. The third is the only durable path for real-time or custom analysis.
Below is each approach, sample chart types, and a 30-line Python example to get a working chart in under an hour.
Tier 1: Free Pre-Built Dashboards
Several free tools surface common Twitter charts without coding:
Trendsmap — geographic heatmap of trending topics, free with daily refresh.
Vidooly Free — channel-level engagement charts, mostly for content creators.
Tweet Binder Free — hashtag campaign reporting, free tier limited to 30-day windows.
These work for casual research. Limitations: can't customize what's charted, can't combine multiple data sources, often capped at recent windows. If your question is "show me top accounts by engagement for hashtag X last week," these will answer. If you need "compute correlation between tweet sentiment and stock price for $TICKER over the past 6 months," you'll need to build.
Tier 2: Public Datasets (Kaggle, archive.org)
Search Kaggle for "Twitter dataset" — thousands of pre-built CSVs covering political events, sports, brand mentions, COVID-era tweets, etc. Sizes range from 10K to 100M+ tweets.
Good for: academic research, machine learning training, historical analysis of completed events.
Limitations: every Kaggle dataset is a snapshot. The original tweets may have been deleted from Twitter; the dataset still has them, but you can't tell what's been altered. Datasets also rarely cover "yesterday" — they're typically months to years old.
Charting these: standard pandas + matplotlib / plotly workflow. Load CSV, group by time / hashtag / account, render.
Tier 3: API-Based Custom Charts
Best path if you need live data or your specific chart type isn't available pre-built.
Workflow:
1. Pull tweets via API into a local store (SQLite or PostgreSQL).
2. Run aggregation queries (group by time bucket, compute engagement, etc.).
3. Push to a charting library (plotly for interactive, matplotlib for static, Grafana / Metabase for dashboards).
Pricing: at $0.15 per 1,000 tweets through third-party APIs, pulling 50K tweets/month for daily-refresh charts costs ~$7.50/month. Compare to $200/month official Basic tier.
Common Chart Types and How to Build Them
Tweet volume over time: group tweets by hour, count, line chart. Tells you the time-of-day rhythm. Useful for picking content publishing slots.
Sentiment over time: classify each tweet (positive / negative / neutral with a lightweight model like VADER or HuggingFace's sentiment pipeline), then plot stacked-area over time. Reveals brand health curves.
Engagement scatter (likes × retweets): plot each tweet as a point. Outliers in the upper-right are viral candidates worth post-mortem study.
Hashtag growth curve: pull tweets matching hashtag, count unique users per day, plot cumulative. Shows adoption velocity.
Follower delta: check account's follower count daily via `user/info`, plot the day-over-day change. Tells you about content effectiveness.
30-Line Starter: Tweet Volume By Hour
Below is a complete working script that pulls last 24h of tweets matching a query and renders an hourly volume chart. Replace `YOUR_API_KEY` and run.
import requests
from datetime import datetime
from collections import defaultdict
import matplotlib.pyplot as plt
API_KEY = "YOUR_API_KEY"
QUERY = "acmecorp lang:en -filter:retweets"
def pull_tweets(query, max_results=500):
out = []
cursor = None
while len(out) < max_results:
params = {"query": query, "limit": 100, "sort": "recent"}
if cursor:
params["cursor"] = cursor
r = requests.get(
"https://api.twitterapi.io/v1/search/advanced",
params=params,
headers={"X-API-Key": API_KEY},
timeout=30,
)
body = r.json()
out.extend(body["data"])
cursor = body.get("next_cursor")
if not cursor:
break
return out
tweets = pull_tweets(QUERY)
by_hour = defaultdict(int)
for t in tweets:
hour = datetime.fromisoformat(t["created_at"].replace("Z", "")).hour
by_hour[hour] += 1
hours = sorted(by_hour.keys())
plt.bar(hours, [by_hour[h] for h in hours])
plt.xlabel("Hour (UTC)")
plt.ylabel("Tweet count")
plt.title(f"Tweets matching '{QUERY}' — last 24h")
plt.savefig("tweet_volume.png")
print("Chart saved to tweet_volume.png")Frequently asked
Where can I find free Twitter datasets to chart?
Kaggle hosts thousands of public Twitter datasets covering politics, sports, brands, COVID-era data, and more. Sizes from 10K to 100M+ tweets. Datasets are static snapshots — for live charts you need API access.
How do I chart real-time Twitter data?
Pull tweets via API into local storage (every 5-15 minutes via cron), then point a charting tool (Grafana, Metabase, Plotly Dash) at your storage. Real-time webhooks let you push to a streaming dashboard with < 1s latency.
What's the easiest way to chart Twitter engagement?
Free dashboards like Vidooly cover engagement for content creators. For custom analysis, pull 30 days of tweets via API ($4-15 cost), load into pandas, group by date, plot likes/retweets/replies. Done in under an hour.
Can I make charts of trending hashtags?
Yes. Poll the trends API every 5-10 minutes, save each trend list snapshot, then plot a hashtag's rank trajectory over time. Useful for measuring campaign velocity or detecting astroturfing patterns.
What charting library is best for Twitter data in Python?
Plotly for interactive dashboards (zoomable, exportable HTML). Matplotlib for static reports / PDFs. Seaborn for quick statistical plots. Grafana if you want a always-on team dashboard that pulls from your database.
Continue reading
Starter credits cover end-to-end testing. No card, no application queue.