I am doing a small project (while trying to educate myself in OCR and data analysis), and I am facing a problem I cannot solve.

I am plotting this graph of speed relative to time, but due to the inconsistency of my OCR results, I am facing the problem of random dots.

def remove_outliers(df: pd.DataFrame) -> pd.DataFrame:
    """
    Change obvious outliers to None in the DataFrame.

    Args:
        df (pd.DataFrame): The DataFrame to clean.

    Returns:
        pd.DataFrame: The cleaned DataFrame.
    """
    # Define thresholds for outliers
    sh_speed_threshold = 6000  # Example threshold for Superheavy speed 
    sh_altitude_threshold = 100  # Example threshold for Superheavy altitude
    ss_speed_threshold = 30000  # Example threshold for Starship speed
    ss_altitude_threshold = 200  # Example threshold for Starship altitude

    # Change outliers to None
    df.loc[df['superheavy_speed'] > sh_speed_threshold, 'superheavy_speed'] = None
    df.loc[df['superheavy_altitude'] > sh_altitude_threshold, 'superheavy_altitude'] = None
    df.loc[df['starship_speed'] > ss_speed_threshold, 'starship_speed'] = None
    df.loc[df['starship_altitude'] > ss_altitude_threshold, 'starship_altitude'] = None

    return df

def delete_outliers_within_window(df: pd.DataFrame, window_size: int = 5) -> pd.DataFrame:
    """
    Delete outliers within a rolling window from the DataFrame.

    Args:
        df (pd.DataFrame): The DataFrame to clean.
        window_size (int): The window size for the rolling operations.

    Returns:
        pd.DataFrame: The cleaned DataFrame.
    """
    for column in ['superheavy_speed', 'superheavy_altitude', 'starship_speed', 'starship_altitude']:
        rolling_median = df[column].rolling(window=window_size, center=True).median()
        rolling_std = df[column].rolling(window=window_size, center=True).std()
        df = df[(df[column] >= (rolling_median - 2 * rolling_std)) & (df[column] <= (rolling_median + 2 * rolling_std))]
    return df

with open(json_path, "r") as f:
    data = json.load(f)
pd.set_option('display.max_columns', None)  # Add this line to display all columns
result, entry = validate_json_structure(data)
if not result:
    print("Invalid JSON structure.", entry)
    
df = pd.DataFrame(data)
df.drop(columns=["time"], inplace=True)  # Drop the "time" column

# Split "superheavy" and "starship" columns into separate "speed" and "altitude" columns with progress bar
for column in tqdm(["superheavy", "starship"], desc="Separating columns"):
    df[[f"{column}_speed", f"{column}_altitude"]] = df[column].apply(pd.Series) 

# Drop the original "superheavy" and "starship" columns
df.drop(columns=["superheavy", "starship"], inplace=True)

# Sort the DataFrame by real time
df.sort_values(by="real_time", inplace=True)

# # # Remove obvious outliers
df = remove_outliers(df)

delete_outliers_within_window(df, window_size=10000)

# Set all Superheavy's data to None after 7 minutes
seven_minutes = 7 * 60  # 7 minutes in seconds
df.loc[df['real_time'] > seven_minutes, ['superheavy_speed', 'superheavy_altitude']] = None

Said graph

Is there any way to remove them using Pandas or any other library? I’ve tried using rolling windows, but it’s not much of a help.