Influence Factors

Factors are the second big data concept besides mood. While mood says "how I'm doing", factors say "what happened during the day". The combination of both is what makes correlations and patterns possible in the first place.

This chapter explains where the suggestions come from, how you create your own factors and categories and how the app automatically detects weekday patterns.

What factors are

Factors are tags that you assign to every entry. "Did sports", "Lots of coffee", "Argument/Conflict", "Meditated", "Cycle week". Each factor belongs to a category like Sleep, Social, Self-care or Movement.

InnerPulse comes with an extensive collection of preinstalled factors, spread across eleven categories. On top of that come your own factors and categories, which you can freely create. Over time, the app learns which factors are important for you and which you rarely set.

The eleven system categories

The preinstalled factors live in the following categories:

• Sleep: Slept well, Slept poorly, Enough sleep (7h+), Little sleep (<6h), Early to bed, Late to bed, Slept through, Woke up often.
• Nutrition: Ate healthy, Ate unhealthy, Cooked myself, Fast food, Drank enough, Lots of coffee, Alcohol, High protein, Skipped a meal.
• Work: Productive, Unproductive, Focused, Unfocused, Deep Work, Topic-hopping, Relaxed day, Deadline stress, Left on time, Overtime.
• Movement: Did sports, No sports, Moved a lot, Only sat, Outdoors/Nature, Running/Jogging, Strength training, Walk, Stretching/Yoga. Plus a larger selection of specific workout factors that activate via the Apple Health integration (cycling, swimming, hiking, HIIT, Pilates, dancing, climbing, tennis, martial arts and more).
• Social: Time with friends, Time with family, Good conversation, Argument/Conflict, Felt lonely, Felt understood, Date/Romance, Met new people, Social occasion.
• Digital: Little screen time, Lots of screen time, No phone before sleep, Phone before sleep, Social media doomscroll, Social media break, Digital detox.
• Entertainment: Watching videos, Series, Movies, Gaming, Shorts, Doomscrolling.
• Self-care: Time off/Me-time, No time for myself, Relaxed, Brooding/Worry, Hobby/Creative, Worked through, Meditated, Journaling, Listened to music, Read.
• Health: Headache, Back pain, Cold/Sick, Allergies, Period cramps, PMS, Cycle week, Took medication, Took supplements, Energetic, Low energy/Tired, Digestive issues.
• Environment: Sunny/Good weather, Gray/Rain, Loud/Restless, Tidy environment, Chaotic environment, Travel/On the road.
• Finances: Saved money, Unplanned expense, Salary/Income, Financial worries.

You can hide system factors but not delete them. That protects against accidental data loss. If an old entry references a factor you no longer see today, the connection stays intact.

Custom categories and factors

In the factor overview, tap New category. Pick name, color and symbol. Within a category, you then create your own factors.

How to rename: You open custom factors for editing by swiping right on the factor in the category detail view. A blue edit field (pencil icon) appears. Tap it and the edit sheet opens, where you can adjust name, symbol and weekday affinity. You rename a custom category itself by opening it and tapping the pencil icon in the top-right toolbar. System factors and system categories are curated and not renamable.

Renaming without data loss: Factors have a stable UUID internally. When you rename a custom factor, the app automatically migrates the new name into all existing entries, into co-occurrence scores and into pair references. If you change "Stress" to "Work stress", the new name also appears in old entries and the entire history stays connected.

Deleting also removes it from old entries: When you delete a custom factor, it is removed from all entries where you had used it at the same time. If you delete a custom category, it takes its contained custom factors with it, and their references are removed from all entries too. If a factor is still in use, InnerPulse asks first and shows how many entries it will be removed from. With zero uses it is deleted without asking. Mood scores, notes and all other factors in the affected entries stay unchanged.

Smart suggestions

When you start a new entry, the likely relevant factors appear at the top. They are not a random selection. InnerPulse computes per factor a score from six dimensions, each with its own weight:

1. Weekday affinity (weight 5.0): If a factor is assigned to a weekday and today is exactly that weekday, it gets the biggest bonus. E.g. "Did sports" on Mondays.
2. Frequency (up to 3.0): Overall frequently used factors stay visible because they are apparently relevant for you.
3. Recency (up to 2.0): Factors you set in the last three entries get a bonus. Yesterday weighs strongest (+2.0), the day before (+1.5), three days ago (+1.0).
4. Time-of-day affinity (up to 1.5): Some factors show up in the morning, others in the evening. This assignment is set per category as default and can be overridden by the user.
5. Mood signal (1.0): If you currently choose a mood value of 4 or lower, negative factors get a bonus. At 7 or higher, positive factors get the bonus. At medium mood (5 or 6) this dimension is neutral.
6. Co-occurrence (up to 1.5): Factors that you often set together with already chosen factors are suggested. If you choose "Argument/Conflict", "Slept poorly" often appears too, because the app has seen this connection in your data.

Co-occurrence scores decay. With every new entry, all existing connections are weighted by factor 0.97. After 10 entries, a co-occurrence score retains about 74 percent of its value, after 50 about 22 percent. That keeps the model current and old, no longer matching patterns fade away.

Factor polarity

Every factor carries a polarity: positive, negative or neutral. It is derived from the factor name (system factors) or learned over time from the mood values (custom factors).

Polarity concretely feeds into the fifth dimension of smart suggestions, the mood signal:

• If you set the mood value in the new entry to 4 or lower, the app assumes the day was stressful. It accordingly prefers to suggest negative factors, so you quickly find what made the day hard (e.g. "Slept poorly", "Argument/Conflict", "Deadline stress").
• If you set the mood value to 7 or higher, the app highlights positive factors (e.g. "Did sports", "Good conversation", "Relaxed").
• At medium mood (5 or 6), polarity is neutral and doesn't influence the order.

It's not about excluding a factor. Negative factors appear further down the list when mood is good, but stay visible. It's just about sort order, so that what you probably need is on top.

Detected patterns

Once enough entries are present, the pattern detection service checks your factors for weekday patterns. If a factor appears in at least 60 percent of entries on a specific weekday, InnerPulse suggests: "You often use 'Did sports' on Mondays. Should I suggest it automatically on Mondays?"

This section only appears in the factor management when the app has actually found a pattern. As long as no anomaly is present, the area is invisible.

You can set the threshold for triggering yourself. By default, the app only suggests a pattern when a factor occurs at least five times on the same weekday. In the factor management you can change this value to 3, 5, 7 or 10, depending on how early or conservative you want suggestions. Lower value: earlier but shakier. Higher value: later but more reliable.

When accepting a suggestion, the weekday affinity is set for the factor, and the smart suggestion engine considers this more strongly in future (dimension 1, weight 5.0).

Base factors and impact factors

In the Analysis two groups appear:

• Base factors are factors that appear in at least 60 percent of your entries. They are your background. "Slept well" or "Time with family" could be base for you. They have low variance because they are almost always there.
• Impact factors are rarer, but when they come, they act clearly. "Argument/Conflict", "Travel/On the road", "Period cramps".

The distinction is statistically important: a factor that occurs in 95 percent of days can't give an impact statement because there are barely any comparison days. A factor that comes ten times and then always causes a mood difference of 2 points is highly relevant.

Auto-activation via Apple Health

If you have Apple Health connected, InnerPulse automatically activates matching factors. The logic with sleep threshold, personal baseline and the learning effect after three rejections is in the chapter Apple Health & Weather. Auto-activated factors are marked with a heart symbol. Per factor, auto-activation can be turned off at any time.

What to know

• The threshold for pattern detection is not fixed. You set it in the factor management yourself (3, 5, 7 or 10 occurrences), default is 5.
• System factors are deliberately curated. They are not arbitrarily extensible because their synonyms and polarities are manually maintained. If something is missing, create it as a custom factor.
• Renaming is safe. The app automatically migrates the new name into all old entries and co-occurrence data.
• The suggestion engine computes all scores at runtime, no background job, no server. With a thin data basis, this takes a few milliseconds; with thousands of entries, imperceptibly longer.
• Setting factors on an entry and understanding them in the analysis are two different tabs. See Mood Journal for setting and Analysis for the correlation analysis.