def __init__(self, json_data_path: Union[str, Path] = None) -> None:
"""
Instantiate, will store the entire json as an attribute, then two pandas DataFrame.
One contains usage data, the other contains messages data.
Args:
json_data_path: string, path (absolute or relative) to json file with your Tinder data.
"""
self.json_file_path: Path = Path(json_data_path).absolute()
with timeit(lambda spanned: logger.debug(f"Loaded json data in {spanned:.4f} seconds")):
with self.json_file_path.open("r") as f:
self.json_data = json.load(f)
with timeit(lambda spanned: logger.debug(f"Processed usage data in {spanned:.4f} seconds")):
self.usage_df: pd.DataFrame = pd.DataFrame.from_dict(self.json_data["Usage"])
self.usage_df.index = pd.to_datetime(self.usage_df.index)
with timeit(
lambda spanned: logger.debug(f"Processed message data in {spanned:.4f} seconds")
):
self.messages_df: pd.DataFrame = pd.DataFrame(self.json_data["Messages"])
self.messages_df.match_id = self.messages_df.match_id.apply(
lambda x: int(x.split(" ")[1])
)
self.messages_df.set_index("match_id", inplace=True)
self.messages_df.sort_index(inplace=True)
logger.debug("Initialisation complete")
def output_usage_statistics(self) -> None:
"""
Print out some statistics about your Tinder usage through the period you've used it.
Returns:
Nothing, will just print.
"""
with timeit(
lambda spanned: logger.debug(f"Gathered usage statistics in {spanned:.4f} seconds")
):
app_opens: pd.Series = self.usage_df.app_opens
creation_date: str = app_opens.index.min()
last_use_date: str = app_opens.index.max()
days_with_use: int = app_opens[app_opens != 0].size
time_period = pd.to_timedelta(last_use_date - creation_date)
print("\n---- Usage Statistics ----")
print(f"Account creation date: {creation_date}")
print(f"Last usage date: {last_use_date}")
print(f"Days spent with an active Tinder account: {time_period.days}")
print(
f"Opened on {days_with_use} unique days, and untouched on "
f"{time_period.days - days_with_use} unique days."
)
print(f"Average app-open per usage-day: {app_opens[app_opens != 0].mean():.1f}\n")
def sankey_metrics(self) -> dict:
"""
Extract and return usage and message metrics from the 'usage_df' and 'messages_df'
dataframes,that will be necessary for the Sankey diagram.
The 'usage_df' dataframe is indexed by date and fairly simple.
The 'messages_df' dataframe is indexed by match ID (starting at 1) and its 'messages'
column is a list in which is a dictionnary for each message sent to the specific match,
with keys 'to', 'from' and 'message'.
The values calculated from 'usage_df' are simply summation over time of columns.
Some insight on keys taken from 'messages_df' and their values' calculation:
contacted: number of rows without an empty list as values.
replied: number of rows with >= 3 dicts (messages sent) in the list. Since Tinder
doesn't give messages sent by your matches, I assume you sending 3 or more
means they've answered.
short_conversations: number of rows with <= 4 dicts (messages sent) in the list.
long-conversations: number of rows with >= 20 dicts (messages sent) in the list.
Returns:
A dictionary with the different usage metrics, all integers.
"""
with timeit(
lambda spanned: logger.debug(
f"Gathered Sankey general metrics in {spanned:.4f} seconds"
)
):
sankey_dict: dict = {
"passes": self.usage_df.swipes_passes.sum(),
"likes": self.usage_df.swipes_likes.sum(),
"swipes": self.usage_df.swipes_passes.sum() + self.usage_df.swipes_likes.sum(),
"matches": self.usage_df.matches.sum(),
"sent_messages": self.usage_df.messages_sent.sum(),
"received_messages": self.usage_df.messages_received.sum(),
}
with timeit(
lambda spanned: logger.debug(
f"Gathered Sankey message metrics in {spanned:.4f} seconds"
)
):
sankey_messages_dict = {
"contacted": self.messages_df.messages.size
- sum(1 for e in self.messages_df.messages if len(e) == 0),
"replied": sum(1 for e in self.messages_df.messages if len(e) >= 3),
"short_conversations": sum(1 for e in self.messages_df.messages if len(e) <= 4),
"long_conversations": sum(1 for e in self.messages_df.messages if len(e) >= 20),
}
return dict(sankey_dict, **sankey_messages_dict)
def output_message_statistics(self) -> None:
"""
Compute and output some statistics on messages from your Tinder data.
Returns:
Nothing, will just print.
"""
with timeit(
lambda spanned: logger.debug(f"Gathered message statistics in {spanned:.4f} seconds")
):
total_messages_sent: int = self.usage_df.messages_sent.sum()
total_messages_received: int = self.usage_df.messages_received.sum()
print("\n---- Message Statistics ----")
print(f"Sent a total of {total_messages_sent} messages")
print(f"Received a total of {total_messages_received} messages\n")
def plot_swipes_weekday_relative_stats(
self, figsize: Tuple[int, int] = (20, 12), showfig: bool = False, savefig: bool = False
):
"""
Plot percentage of the total number of swipes that were likes, passes and those that
resulted in matches, on a monthly basis.
Args:
figsize: figure size
showfig: if this is set to True, the figure will be shown. Defaults to False.
savefig: if this is set to True, the figure will be saved. Defaults to False.
Returns:
Nothing, just plots.
"""
logger.debug("Plotting swipes weekday relative insights")
with timeit(
lambda spanned: logger.debug(
f"Gathered relative weekday swipe statistics in {spanned:.4f} seconds"
)
):
# Careful, index sorts alphabetically for now)
vals_weekday = self.usage_df.groupby(self.usage_df.index.weekday_name).sum()
vals_weekday.fillna(0, inplace=True)
# Getting index as categorical properly ordered weekdays
weekday_categories = pd.api.types.CategoricalDtype(categories=WEEK_DAYS, ordered=True)
vals_weekday.index = vals_weekday.index.astype(weekday_categories)
vals_weekday["total_swipes"] = vals_weekday.swipes_likes + vals_weekday.swipes_passes
vals_weekday["likes_ratio"] = (
100 * vals_weekday.swipes_likes / vals_weekday.total_swipes
)
vals_weekday["matches_ratio"] = 100 * vals_weekday.matches / vals_weekday.total_swipes
vals_weekday["passes_ratio"] = (
100 * vals_weekday.swipes_passes / vals_weekday.total_swipes
)
vals_weekday.fillna(0, inplace=True)
fig, axis1 = plt.subplots(figsize=figsize)
axis1.plot(
vals_weekday.index,
vals_weekday.likes_ratio,
color="blue",
label="'Like' Swipes",
lw=1.5,
)
axis1.plot(
vals_weekday.index,
vals_weekday.passes_ratio,
color="red",
label="'Pass' Swipes",
lw=1.5,
)
axis1.plot(
vals_weekday.index, vals_weekday.matches_ratio, color="green", label="Matches", lw=1.5
)
axis1.set_xlabel("Date")
axis1.set_ylabel("Percentage of Total Swipes [%]")
axis1.set_title("Monthly Ratio of Likes, Passes and Matches to Total Swipes")
axis1.tick_params(axis="x", rotation=75)
axis1.legend()
fig.tight_layout()
if showfig:
logger.debug("Showing swipes weekday relative stats figure")
plt.show()
if savefig:
logger.debug("Saving swipes weekday relative stats figure")
fig.savefig("plots/swipes_weekdays_relative_stats.png", format="png", dpi=500)
logger.success(
"Saved swipes weekdays relative stats plot as "
"'plots/swipes_weekdays_relative_stats.png'"
)
def plot_swipes_weekday_stats(
self, figsize: Tuple[int, int] = (20, 12), showfig: bool = False, savefig: bool = False
) -> None:
"""
Compute the number of right and left swipes sent for each day of the week, then output
it as a stacked barplot. Also plot the number of matches for each weekday.
Args:
figsize: figure size
showfig: if this is set to True, the figure will be shown. Defaults to False.
savefig: if this is set to True, the figure will be saved. Defaults to False.
Returns:
Nothing, just plots.
"""
logger.debug("Plotting swipes weekday insights")
with timeit(
lambda spanned: logger.debug(
f"Gathered weekday swipe statistics in {spanned:.4f} seconds"
)
):
# Careful, index sorts alphabetically for now)
vals_weekday = self.usage_df.groupby(self.usage_df.index.weekday_name).sum()
vals_weekday.fillna(0, inplace=True)
# Getting index as categorical properly ordered weekdays
weekday_categories = pd.api.types.CategoricalDtype(categories=WEEK_DAYS, ordered=True)
vals_weekday.index = vals_weekday.index.astype(weekday_categories)
vals_weekday.sort_index(inplace=True)
fig, axis1 = plt.subplots(figsize=figsize)
axis2 = axis1.twinx()
axis1.bar(
vals_weekday.index,
vals_weekday.swipes_likes,
label="'Like' Swipes",
color="blue",
alpha=0.4,
)
axis1.bar(
vals_weekday.index,
vals_weekday.swipes_passes,
bottom=vals_weekday.swipes_likes,
label="'Pass' Swipes",
color="red",
alpha=0.4,
)
axis2.plot(vals_weekday.index, vals_weekday.matches, color="grey", lw=1.5)
axis1.set_xlabel("Date")
axis1.set_ylabel("Number of Swipes")
axis1.set_title("Swipes per Weekday")
axis1.tick_params(axis="x", rotation=75)
axis1.legend()
axis2.set_ylabel("Number of Matches", color="grey")
axis2.tick_params(axis="y", labelcolor="grey")
axis2.grid(False)
fig.tight_layout()
if showfig:
logger.debug("Showing swipes weekday stats figure")
plt.show()
if savefig:
logger.debug("Saving swipes weekday stats figure")
fig.savefig("plots/swipes_weekdays_stats.png", format="png", dpi=500)
logger.success("Saved swipes weekdays stats plot as 'plots/swipes_weekdays_stats.png'")
def plot_swipes_monthly_stats(
self, figsize: Tuple[int, int] = (20, 12), showfig: bool = False, savefig: bool = False
):
"""
Compute the monthly left and right swipes, then output it as a stacked barplot. Also plot
the monthly number of matches.
Args:
figsize: figure size
showfig: if this is set to True, the figure will be shown. Defaults to False.
savefig: if this is set to True, the figure will be saved. Defaults to False.
Returns:
Nothing, just plots.
"""
logger.debug("Plotting swipes monthly insights")
with timeit(
lambda spanned: logger.debug(
f"Gathered monthly swipe statistics in {spanned:.4f} seconds"
)
):
vals_monthly = self.usage_df.groupby(pd.Grouper(freq="M")).sum()
vals_monthly.fillna(0, inplace=True)
fig, axis1 = plt.subplots(figsize=figsize)
axis2 = axis1.twinx()
axis1.bar(
vals_monthly.index,
vals_monthly.swipes_likes,
width=1,
label="'Like' Swipes",
color="blue",
alpha=0.4,
)
axis1.bar(
vals_monthly.index,
vals_monthly.swipes_passes,
bottom=vals_monthly.swipes_likes,
width=1,
label="'Pass' Likes",
color="red",
alpha=0.4,
)
axis2.plot(vals_monthly.index, vals_monthly.matches, color="grey", label="Matches", lw=1.5)
axis1.set_xlabel("Date")
axis1.set_ylabel("Number of Swipes")
axis1.set_title("Swipes per Month")
axis1.tick_params(axis="x", rotation=75)
axis2.set_ylabel("Number of Matches", color="grey")
axis2.tick_params(axis="y", labelcolor="grey")
axis2.grid(False)
fig.legend(loc=2, bbox_to_anchor=(0, 1), bbox_transform=axis1.transAxes)
fig.tight_layout()
if showfig:
logger.debug("Showing swipes monthly stats figure")
plt.show()
if savefig:
logger.debug("Saving swipes monthly stats figure")
fig.savefig("plots/swipes_monthly_stats.png", format="png", dpi=500)
logger.success("Saved swipes monthly stats plot as 'plots/swipes_monthly_stats.png'")
def plot_messages_monthly_stats(
self, figsize: Tuple[int, int] = (20, 12), showfig: bool = False, savefig: bool = False
) -> None:
"""
Compute the monthly sent and received number of messages, then output it as a stacked
barplot. Also plot the % of received/sent ratio.
Args:
figsize: figure size
showfig: if this is set to True, the figure will be shown. Defaults to False.
savefig: if this is set to True, the figure will be saved. Defaults to False.
Returns:
Nothing, just plots.
"""
logger.debug("Plotting messages monthly insights")
with timeit(
lambda spanned: logger.debug(
f"Gathered monthly message statistics in {spanned:.4f} seconds"
)
):
vals_monthly = self.usage_df.groupby(pd.Grouper(freq="M")).sum()
vals_monthly["ratio"] = (
100 * vals_monthly.messages_received / vals_monthly.messages_sent
)
vals_monthly.fillna(0, inplace=True)
fig, axis1 = plt.subplots(figsize=figsize)
axis2 = axis1.twinx()
axis1.bar(
vals_monthly.index,
vals_monthly.messages_sent,
width=1,
label="Messages Sent",
color="blue",
alpha=0.4,
)
axis1.bar(
vals_monthly.index,
vals_monthly.messages_received,
bottom=vals_monthly.messages_sent,
width=1,
label="Messages Received",
color="red",
alpha=0.4,
)
axis2.plot(vals_monthly.index, vals_monthly.ratio, color="grey", lw=1.5)
axis1.set_xlabel("Date")
axis1.set_ylabel("Number of Messages")
axis1.set_title("Messages per Month")
axis1.tick_params(axis="x", rotation=75)
axis1.legend()
axis2.set_ylabel("Received/Sent [%]", color="grey")
axis2.tick_params(axis="y", labelcolor="grey")
axis2.grid(False)
fig.tight_layout()
if showfig:
logger.debug("Showing message monthly stats figure")
plt.show()
if savefig:
logger.debug("Saving message monthly stats figure")
fig.savefig("plots/messages_monthly_stats.png", format="png", dpi=500)
logger.success("Saved message monthly stats plot as 'plots/messages_monthly_stats.png'")
