Read Meta Data File w/o File Extension

Description

Read a meta file without extension -> unknown format Tries to determine format (yaml, json) and return the metadata path must point to the file without extension e.g. swissdata_wd/set_id/set_id

Usage

.read_swissdata_meta_unknown_format(path)

Arguments

Value

Meta list if file could be located, empty list otherwise

CH GDP Growth Contributions

Description

A list of time series containing sector contributions to Swiss GDP over time.

Usage

CHGDP

Format

List list of six time series of class ts, containing contributions to Swiss GDP growth

manufacturing

Growth contribution of manufacturing.

energy

Growth contribution of energy, water sector

construction

Growth contribution construction sector.

hotels

Growth contribution of hotels.

fin_insur

Growth contribution of financial services and insurances.

other

Growth contribution of other sectors.

Source

https://www.seco.admin.ch/seco/en/home/wirtschaftslage—wirtschaftspolitik/Wirtschaftslage/bip-quartalsschaetzungen-/daten.html

KOF Barometer - Swiss Business Cycle Indicator

Description

A list of time series containing two time series the KOF Barometer and the growth of Swiss GDP over time. KOF Barometer is a monthly business cycle indicator computed by the KOF Swiss Economic Institute. The GDP growth rate is used as a reference series to the Barometer.

Usage

KOF

Format

A list of two time series of class ts

kofbarometer

KOF Barometer Indicator

'

reference

Reference series to KOF Barometer, change in Swiss GDP compared to previous month

baro_point_fc

Auto Arima point forecast of the KOF Barometer

baro_lo_80

Auto Arima 80 percent CI lower bound of the KOF Barometer forecast

baro_hi_80

Auto Arima 80 percent CI upper bound of the KOF Barometer forecast

baro_lo_95

Auto Arima 95 percent CI lower bound of the KOF Barometer forecast

baro_hi_95

Auto Arima 95 percent CI upper bound of the KOF Barometer forecast

...

Source

https://kof.ethz.ch/en/forecasts-and-indicators/indicators/kof-economic-barometer.html

Provide Colorblind Compliant Colors

Description

8 Hex RGB color defintions suitable for charts for colorblind people.

Usage

color_blind()

Compute Decimal Time from a ts Period Vector

Description

Standard ts object use a vector of length two to store a period. E.g. 2010,1 means first quarter of 2010, if the series was quarterly and first month if the series was monthly etc.

Usage

compute_decimal_time(v, f)

Arguments

Concatenate to Non-Overlapping Time Series

Description

Append one time series to another. This only works for non-overlapping time series of the same frequency. For overlapping time series please see resolveOverlap.

Usage

concat_ts(ts1, ts2)

Arguments

Create an Overview data.table of (last) observations

Description

Create a data.table that shows the i-th obsersvation of several time series.

Usage

create_cross_sec_overview(list_of_rows, col_labels, tsl, selected_period)

Arguments

Examples

tsl <- generate_random_ts(10, lengths = 20)
list_of_rows <- list(
  "group 1" = c("ts1", "ts2", "ts3", "ts4"),
  "group 2" = c("ts5", "ts6", "ts7", "ts10")
)
# These are no real +,=,- values just random data.
create_cross_sec_overview(
  list_of_rows,
  c("+", "=", "-", "random"),
  tsl, c(1988, 12)
)

Flexible Function to Create Time Series Dummy Variables

Description

Generate time series with a default value that is changed within a certain subperiod. The function allows for additional convenience when specifying single period dummies and dummies that go from a certain point in time to the end of the series.

Usage

create_dummy_ts(
  end_basic,
  dummy_start,
  dummy_end = NULL,
  sp = T,
  start_basic = c(1980, 1),
  basic_value = 0,
  dummy_value = 1,
  frequency = 4
)

Arguments

Author(s)

Matthias Bannert

Turn data.frame to Regular Monthly or Quarterly Time Series

Description

Turn a data.frame with date columns to a regular time series object if possible. Design to work with quarterly and monthly data.

Usage

df_to_reg_ts(
  dframe,
  var_cols,
  year_col = "year",
  period_col = "month",
  freq = 12,
  return_ts = T,
  by = NULL
)

Arguments

Examples

start_m <- as.Date("2017-01-01")
df_missing <- data.frame(
  date = seq(start_m, by = "2 months", length = 6),
  value = 1:6,
  another_value = letters[1:6],
  yet_another_col = letters[6:1]
)
df_to_reg_ts(df_missing, c("value", "another_value"))
df_to_reg_ts(df_missing, c("value", "another_value"), return_ts = FALSE)

Fill Up a Time Series with NAs

Description

When plotting a time series you might want set the range of the plot a little wider than just the start and end date of the original series. This function add fills up the current period (typically year) with NA.

Usage

fill_year_with_nas(x, add_periods = 1, fill_up_start = FALSE)

Arguments

Generate a list of random time series

Description

Useful for development or generating easily reproducible examples

Usage

generate_random_ts(
  n = 1,
  lengths = 36,
  starts = 1988,
  frequencies = 12,
  ranges_min = -1,
  ranges_max = 1,
  shifts = 0,
  ts_names = sprintf("ts%d", 1:n),
  seed = 30042018,
  random_NAs = FALSE,
  random_NA_proportions = 0.1,
  normally_distributed = FALSE,
  normal_means = 0,
  normal_sds = 1,
  frequency_shifts = FALSE,
  frequency_shift_after = 0.5
)

Arguments

Details

Except for n and ts_names, all parameters accept either a single value or a vector of values. If a single value is supplied, that value is used for all time series being generated. If a vector is supplied, its values will be used for the corresponding series (e.g. starts[1] is used for the first series, starts[2] for the second and so on). Vectors are recycled if n is larger than their length.

If a ts_names vector is supplied, it must have length n and must not contain duplicates.

Value

A list of ts objects

Examples

generate_random_ts()

generate_random_ts(n = 3, ranges_min = c(-10, 0, 10), ranges_max = 20, starts = 2011)

Helper to calculate ci colors for legends

Description

Helper to calculate ci colors for legends

Usage

getCiLegendColors(color, n = 1, alpha = NULL)

Arguments

Details

Color may be specified as either a named color or a hex value Transparency may be specified as a hex value, number 0-255 or number 0-1

Value

A vector of non-transparent colors that result from oberlaying color over pure white 1:n times

Compute the Period Vector representation of a Decimal Time value

Description

The period value will be rounded down to the nearest integer. This function is not vectorized so only a single value can be converted at a time.

Usage

get_date_vector(dtime, frq)

Arguments

Initiate Default Theme

Description

The tsplot methods provide a theme argument which is used to pass on a plethora of useful defaults. These defaults are essentially stored in a list. Sometimes the user may want to tweak some of these defaults while keeping most of them. Hence the init_tsplot_theme function create a fresh list object containing default values for lot of different layout parameters etc. By replacing single elements of the list and passing the entire list to the plot function, single aspects can be tweaked while keeping most defaults. Init defaultTheme does not need any parameters.

This function provides sensible defaults for margins, font size, line width etc. scaled to the dimensions of the output file.

Usage

init_tsplot_theme(
  auto_bottom_margin = FALSE,
  band_fill_color = c(ETH_Petrol = colors$ETH_Petrol$`100`, ETH_Petrol_60 =
    colors$ETH_Petrol$`60`, ETH_Petrol_40 = colors$ETH_Petrol$`40`, ETH_Petrol_20 =
    colors$ETH_Petrol$`20`, ETH_Purple = colors$ETH_Purple$`100`, ETH_Purple_60 =
    colors$ETH_Purple$`60`, ETH_Purple_40 = colors$ETH_Purple$`40`),
  bar_border = "#000000",
  bar_border_lwd = 1,
  bar_fill_color = c(ETH_Petrol = colors$ETH_Petrol$`100`, ETH_Petrol_60 =
    colors$ETH_Petrol$`60`, ETH_Petrol_40 = colors$ETH_Petrol$`40`, ETH_Petrol_20 =
    colors$ETH_Petrol$`20`, ETH_Purple = colors$ETH_Purple$`100`, ETH_Purple_60 =
    colors$ETH_Purple$`60`, ETH_Purple_40 = colors$ETH_Purple$`40`),
  bar_gap = 15,
  bar_group_gap = 30,
  ci_alpha = "44",
  ci_colors = line_colors,
  ci_legend_label = "%ci_value%% ci for %series%",
  default_bottom_margin = 15,
  fill_up_start = FALSE,
  fill_year_with_nas = TRUE,
  highlight_color = colors$ETH_Grey$`20`,
  highlight_window = FALSE,
  highlight_window_end = NA,
  highlight_window_freq = 4,
  highlight_window_start = NA,
  highlight_y_values = NA,
  highlight_y_lwd = 2,
  highlight_y_color = "#000000",
  label_pos = "mid",
  legend_all_left = FALSE,
  legend_box_size = 2,
  legend_col = 1,
  legend_font_size = 1,
  legend_intersp_x = 1,
  legend_intersp_y = 1,
  legend_margin_bottom = 5,
  legend_margin_top = 12,
  legend_seg.len = 2,
  line_colors = c(ETH_Green_60 = colors$ETH_Green$`60`, ETH_Green_100 =
    colors$ETH_Green$`100`, ETH_Petrol_20 = colors$ETH_Petrol$`20`, ETH_Purple_60 =
    colors$ETH_Purple$`60`, ETH_Petrol_60 = colors$ETH_Petrol$`60`, ETH_Purple_100 =
    colors$ETH_Purple$`100`, ETH_Petrol_100 = colors$ETH_Petrol$`100`),
  line_to_middle = TRUE,
  lty = 1,
  lwd = c(2, 3, 1, 4, 2, 4),
  lwd_box = 1.5,
  lwd_quarterly_ticks = 1,
  lwd_x_axis = 1.5,
  lwd_y_axis = 1.5,
  lwd_y_ticks = 1.5,
  lwd_yearly_ticks = 1.5,
  margins = c(NA, 7, 12, 7),
  NA_continue_line = FALSE,
  output_wide = FALSE,
  point_symbol = 1:18,
  pointsize = 12,
  preferred_y_gap_sizes = c(25, 20, 15, 10, 5, 2.5, 1, 0.5),
  quarterly_ticks = TRUE,
  range_must_not_cross_zero = TRUE,
  show_left_y_axis = TRUE,
  show_points = FALSE,
  show_right_y_axis = TRUE,
  show_x_axis = TRUE,
  show_y_grids = TRUE,
  subtitle_adj = 0,
  subtitle_adj_r = 0.9,
  subtitle_cex = 1,
  subtitle_margin = 2,
  subtitle_outer = FALSE,
  subtitle_transform = "toupper",
  sum_as_line = FALSE,
  sum_legend = "sum",
  sum_line_color = c(ETH_Petrol_100 = colors$ETH_Petrol$`100`),
  sum_line_lty = 1,
  sum_line_lwd = 3,
  tcl_quarterly_ticks = -0.4,
  tcl_y_ticks = -0.75,
  tcl_yearly_ticks = -0.75,
  title_adj = 0,
  title_cex.main = 1,
  title_margin = 5,
  title_outer = FALSE,
  title_transform = NA,
  total_bar_margin_pct = 0.2,
  use_bar_gap_in_groups = FALSE,
  use_box = FALSE,
  x_tick_dt = 1,
  xaxs = "i",
  y_grid_color = colors$ETH_Grey$`40`,
  y_grid_count = c(5, 6, 8, 10),
  y_grid_count_strict = FALSE,
  y_las = 2,
  y_range_min_size = NULL,
  y_tick_force_integers = FALSE,
  y_tick_margin = 0.15,
  yaxs = "i",
  yearly_ticks = TRUE
)

init_tsplot_print_theme(
  output_wide = FALSE,
  margins = c(NA, 10/if (output_wide) 1 + 1/3 else 1, 10, 7/if (output_wide) 1 + 1/3 else
    1),
  lwd = scale_theme_param_for_print(c(2, 3, 1, 4, 2, 4), if (output_wide) c(10 + 2/3, 6)
    else c(8, 6)),
  sum_line_lwd = scale_theme_param_for_print(3, if (output_wide) c(10 + 2/3, 6) else c(8,
    6)),
  lwd_box = scale_theme_param_for_print(1.5, if (output_wide) c(10 + 2/3, 6) else c(8,
    6)),
  lwd_x_axis = scale_theme_param_for_print(1.5, if (output_wide) c(10 + 2/3, 6) else c(8,
    6)),
  lwd_yearly_ticks = scale_theme_param_for_print(1.5, if (output_wide) c(10 + 2/3, 6)
    else c(8, 6)),
  lwd_quarterly_ticks = scale_theme_param_for_print(1, if (output_wide) c(10 + 2/3, 6)
    else c(8, 6)),
  lwd_y_axis = scale_theme_param_for_print(1.5, if (output_wide) c(10 + 2/3, 6) else c(8,
    6)),
  lwd_y_ticks = scale_theme_param_for_print(1.5, if (output_wide) c(10 + 2/3, 6) else
    c(8, 6)),
  legend_intersp_y = scale_theme_param_for_print(1, if (output_wide) c(10 + 2/3, 6) else
    c(8, 6)),
  legend_box_size = scale_theme_param_for_print(2, if (output_wide) c(10 + 2/3, 6) else
    c(8, 6)),
  legend_margin_top = 8,
  legend_margin_bottom = 3,
  legend_seg.len = scale_theme_param_for_print(2, if (output_wide) c(10 + 2/3, 6) else
    c(8, 6)),
  pointsize = scale_theme_param_for_print(12, if (output_wide) c(10 + 2/3, 6) else c(8,
    6)),
  ...
)

Arguments

Details

Themes are essentially list that contain par parameters. Below all items are listed, some of them with comments. The per-line parameters (line_colors, lwd, lty, show_points, point_symbol) are recycled if more time series than elements on the corresponding theme vectors are supplied. e.g. if four time series are plotted but only two line_colors are supplied, the first and third series have the first color, while the second and fourth series have the second color. The list contains the following elements:

Author(s)

Matthias Bannert

Examples

## Not run: 
# create a list
data(KOF)
tt <- init_tsplot_theme()
# adjust a single element
tt$highlight_window <- TRUE
# pass the list to tsplot
tsplot(KOF$kofbarometer, theme = tt)
# for more theme examples check the vignette
vignette("tstools")

## End(Not run)

Transform a long format data.frame of time series to a tslist

Description

The data.frame must have three columns "date", "value" and "series" (identifying the time series)

Usage

long_to_ts(
  data,
  keep_last_freq_only = FALSE,
  force_xts = FALSE,
  strip_nas = TRUE
)

Arguments

Turn monthly series with regular NAs to quarter

Description

Monthly series with NAs in non-quarter months are turned to quarterly series. Series without NAs are just returned.

Usage

m_to_q(series)

Arguments

Concat Time Series list wise

Description

Concat overlapping time series list wise. List needs to be of same length. Takes names of list B.

Usage

overlap_sorted_ts_lists(listA, listB)

Arguments

Resolve Overlap Listwise, helpful with SA

Description

Resolve Overlap Listwise, helpful with SA

Usage

overlap_ts_lists_by_name(listA, listB, chunkA = "_f4", chunkB = "_f12")

Arguments

Read data generated by the Swissdata project

Description

Read data from swissdata compliant .csv files and turn them into a list of time series.

Usage

read_swissdata(
  path,
  key_columns = NULL,
  filter = NULL,
  aggregates = NULL,
  keep_last_freq_only = FALSE
)

Arguments

Details

The order of dimensions in key_columns determines their order in the key The resulting ts_key will be of the form <swissdata-set-name>.<instance of key_columns[1]>...

Examples

ds_location <- system.file("example_data/ch.seco.css.csv", package = "tstools")
tslist <- read_swissdata(ds_location, "idx_type")
tsplot(tslist[1])

Read swissdata style yaml timeseries metadata

Description

read_swissdata_meta reads the given .yaml file and converts it into a per-timeseries format.

Usage

read_swissdata_meta(path, locale = "de", as_list = FALSE)

Arguments

Details

If as_list is set to TRUE, the function returns a nested list with one element per timeseries, otherwise a data.table with one row per series.

Import time series data from a file.

Description

If importing from a zip file, the archive should contain a single file with the extension .csv, .xlsx or .json.

Usage

read_ts(
  file,
  format = c("csv", "xlsx", "json", "zip"),
  sep = ",",
  skip = 0,
  column_names = c("date", "value", "series"),
  keep_last_freq_only = FALSE,
  force_xts = FALSE
)

Arguments

Value

A named list of ts objects

Turn an Irregular Time Series to a Regular, ts-Based Series

Description

Adds missing values to turn an irregular time series into a regular one. This function is currently experimental. Only works or target frequencies 1,2,4,12.

Usage

regularize(x)

Arguments

Examples

ts1 <- rnorm(5)
dv <- c(
  seq(as.Date("2010-01-01"), length = 3, by = "3 years"),
  seq(as.Date("2018-01-01"), length = 2, by = "2 years")
)
library(zoo)
xx <- zoo(ts1, dv)
regularize(xx)

dv2 <- c(seq(as.Date("2010-01-01"), length = 20, by = "1 months"))
dv2 <- dv2[c(1:10, 14:20)]
xx2 <- zoo(rnorm(length(dv2)), dv2)
regularize(xx2)

Concatenate Time Series and Resolve Overlap Automatically

Description

Append time series to each other. Resolve overlap determines which of two ts class time series is reaching further and arranges the two series into first and second series accordingly. Both time series are concatenated to one if both series had the same frequency. Typically this function is used concatenate two series that have a certain overlap, but one series clearly starts earlier while the other lasts longer. If one series starts earlier and stops later, all elements of the shorter series will be inserted into the larger series, i.e. elements of the smaller series will replace the elements of the longer series. Usually ts2 is kept.

Usage

resolve_ts_overlap(ts1, ts2, keep_ts2 = T, tolerance = 0.001)

Arguments

Examples

ts1 <- ts(rnorm(100), start = c(1990, 1), frequency = 4)
ts2 <- ts(1:18, start = c(2000, 1), frequency = 4)
resolve_ts_overlap(ts1, ts2)

# automatical detection of correction sequence!
ts1 <- ts(rnorm(90), start = c(1990, 1), frequency = 4)
ts2 <- ts(1:60, start = c(2000, 1), frequency = 4)
resolve_ts_overlap(ts1, ts2)

# both series are of the same length use sequence of arguments.
ts1 <- ts(rnorm(100), start = c(1990, 1), frequency = 4)
ts2 <- ts(1:48, start = c(2003, 1), frequency = 4)
resolve_ts_overlap(ts1, ts2)
ts1 <- ts(rnorm(101), start = c(1990, 1), frequency = 4)
ts2 <- ts(1:61, start = c(2000, 1), frequency = 4)
resolve_ts_overlap(ts1, ts2)
#' clearly dominatn ts2 series
ts1 <- ts(rnorm(50), start = c(1990, 1), frequency = 4)
ts2 <- ts(1:100, start = c(1990, 1), frequency = 4)
resolve_ts_overlap(ts1, ts2)

Set Periods to NA

Description

This function is typically used to discard information in non-quarter month. I.e., data is only kept in January, April, July and December and otherwise set to NA. In combination with m_to_q this function is useful to turn monthly series into quarterly series by letting the quarter month values represent the entire quarter. This can be useful when data was interpolated because of mixing data of different frequencies and needs to be converted back to a regular, quarterly time series.

Usage

set_month_to_NA(series, keep_month = c(1, 4, 7, 10))

Arguments

Examples

tsq <- ts(1:20, start = c(1990, 1), frequency = 4)
aa <- tsqm(tsq)
m_to_q(set_month_to_NA(aa))

Start a Time Series after the Last Internal NA

Description

Internal NAs can cause trouble for time series operations such as X-13-ARIMA SEATS seasonal adjustment. Often, internal NAs only occur at at the beginning of a time series. Thus an easy solution to the problem is to discard the initial part of the data which contains the NA values. This way only a small part of the information is lost as opposed to not being able to seasonally adjust an entire series.

Usage

start_ts_after_internal_nas(series)

Arguments

See Also

stripLeadingNAsFromTs, stripTrailingNAsFromTs

Examples

ts1 <- 1:30
ts1[c(3, 6)] <- NA
ts1 <- ts(ts1, start = c(2000, 1), frequency = 4)
start_ts_after_internal_nas(ts1)

Strip Leading / Trailing NAs from a Time Series Object

Description

Removes NAs to begin with and starts time series index at the first non-NA value.

Usage

strip_ts_of_leading_nas(s)

strip_ts_of_trailing_nas(s)

Arguments

Plot Time Series

Description

Conveniently plot time series.

Usage

tsplot(
  ...,
  tsr = NULL,
  ci = NULL,
  left_as_bar = FALSE,
  group_bar_chart = FALSE,
  relative_bar_chart = FALSE,
  left_as_band = FALSE,
  plot_title = NULL,
  plot_subtitle = NULL,
  plot_subtitle_r = NULL,
  find_ticks_function = "findTicks",
  overall_xlim = NULL,
  overall_ylim = NULL,
  manual_date_ticks = NULL,
  manual_value_ticks_l = NULL,
  manual_value_ticks_r = NULL,
  manual_ticks_x = NULL,
  theme = NULL,
  quiet = TRUE,
  auto_legend = TRUE,
  output_format = "plot",
  filename = "tsplot",
  close_graphics_device = TRUE
)

Arguments

Details

The ci parameter is a 3-level list of the form list( ts1 = list( ci_value_1 = list( ub = upper_bound_ts_object, lb = lower_bound_ts_object ), ... ), ... )

See vignette("tstools") for details.

Interpolate quarterly time series into monthly

Description

Repeat quarterly variables two times to generate a monthly variable.

Usage

tsqm(qts)

Arguments

Examples

tsq <- ts(1:20, start = c(1990, 1), frequency = 4)
tsqm(tsq)

Deprecated function(s) in tstools

Description

These functions are provided for compatibility with older version of the tstools package. They may eventually be completely removed.

Arguments

Details

Transform a wide format data.frame into a tslist

Description

The time series in the data.frame may be stored either rowwise or columnswise. The identifying column must be called date (for columnwise) or series (for rowwise)

Usage

wide_to_ts(data, keep_last_freq_only = FALSE, force_xts = FALSE)

Arguments

Export a list of time series to a file.

Description

Export a list of time series to a file.

Usage

write_ts(
  tl,
  fname = NULL,
  format = "csv",
  date_format = NULL,
  timestamp_to_fn = FALSE,
  round_digits = NULL,
  rdata_varname = "tslist",
  ...
)

Arguments

Details

Additional arguments covered by ...